High-output, low-duty cycle LED strobe circuit

[mj]

I'm looking for ideas on how to make an LED flash so brightly at a low
duty cycle that it's reasonably bright--maybe even close to what it
would be if it were on DC.

I'm building a project where I need to flash white LEDs very brightly
30-50 times a second at about a 0.4% duty cycle. (I'm strobing a
spinning disk, and want to freeze images near the LED--too high a duty
cycle, and the image blurs.) 0.4% is not much time for an LED to be
on. I've heard that you can drive LEDs to up to 10x their normal
forward current without damage (though I guess lifetime is shortened)
if you keep duty cycle to <= 1%.

To try to get a bright enough flash, I got some 0.5W white LEDs that
can take a max DC forward current of 150 mA, and have about a ~3.6
forward drop, producing an intensity of 130k mcd. (Not too clear on
the mcd part.) And dem suckers is bright when you're pumping even 100
mA through them. Like squint-to-look-at-it bright. They're in a
standard 5mm package (T-1 3/4, what is up with that package name?),
though it's sturdier than most you've probably seen. The LED looks
like it's been lifting weights, and the leads are shorter and fatter.

Anyway, I have two 2n2222's hooked up as a Darlington, with +5 Vcc,
driving the front Q's base with ~15mA (with a microcontroller pin).
There's NO current-limiting resistor on the back end, where the second
transistor's collector is attached to +5v, and the emitter goes
through the LED to ground.

Since the LED is only on 0.4% of the time, max, it still simply isn't
very bright. The strobe works--I can see the frozen image on the
spinning disk--but the light is simply anemic.

So, I'm wondering if anyone here knows how to design a circuit that
can dump an amp and a half through an LED for, say, 200 microseconds
at a time or less, at 20-50 Hz. Or if anyone else has ideas about how
to make LEDs look bright even if they're only on half a percent of the
time.

I'm hoping I won't have to dump the LED idea and go with tiny xenon
strobes--not really into figuring how to design a 50Hz photo strobe at
the moment.

Thanks for anyone that wants to provide some ideas.

 

[Jan Panteltje]

I'm looking for ideas on how to make an LED flash so brightly at a low
duty cycle that it's reasonably bright--maybe even close to what it
would be if it were on DC.

I'm building a project where I need to flash white LEDs very brightly
30-50 times a second at about a 0.4% duty cycle. (I'm strobing a
spinning disk, and want to freeze images near the LED--too high a duty
cycle, and the image blurs.) 0.4% is not much time for an LED to be
on. I've heard that you can drive LEDs to up to 10x their normal
forward current without damage (though I guess lifetime is shortened)
if you keep duty cycle to <= 1%.

I have exactly that here, made it for strobe tests.
White LED too, uses an 8 pin PIC, and 2 transistors to get some high current.
Want the asm source? Could perhaps still find it.

 

[Tim Williams]

I have exactly that here, made it for strobe tests.
White LED too, uses an 8 pin PIC, and 2 transistors to get some high current.
Want the asm source? Could perhaps still find it.

A PIC for blinking an LED? Not even LED*s*, but *an* LED?

You are scum among engineers!

(Obligatory ;-) included, since *I hope* you chose a PIC for timing
reasons or something.)

Tim

 

[mj]

A PIC for blinking an LED?  Not even LED*s*, but *an* LED?

You are scum among engineers!

(Obligatory ;-) included, since *I hope* you chose a PIC for timing
reasons or something.)

Tim

Hah! Now I must defend my honor! (Though I must admit to being scum in
other contexts.)

Actually, I'm actually doing this for a dozen LEDs at once. A rotating
transparent disk has [A-Z0-9 ] printed just inside the circumference
around the circle. A motor spins the disk at 20-50 RPM (controlled by
the MCU with PWM). An IR photo-Q watches for a black mark on the
bottom of the disk, and through a Schmitt trigger, interrupts the
processor once per rev. The 12 LEDs are arranged radially below the
row of letters flashing by, and since the MCU knows where the disk is
at all times, it can flash to display a specific character at a
specific position. Shifting the character positions lets you scroll a
character at a time, slight delay or advance on the clock gives smooth
scrolling, alternating between one character and another provides
dissolve, etc. Plus another photo-Q will receive text messages from a
tiny IR keyboard (Targus). I may also network-enable it so it can show
RSS feeds.

Kind of like a POV toy, but strobing spinning characters instead of
drawing the characters with finely-spaced LEDs.

Flashing a single LED, indeed! Harrumph!

:{)

 

[mj]

I have exactly that here, made it for strobe tests.
White LED too, uses an 8 pin PIC, and 2 transistors to get some high current.
Want the asm source? Could perhaps still find it.

Actually, it's the circuit to pump current through those LEDs that I
need. My code works already. It works, but the light is feeble. The
asm source wouldn't do me much good, since I'm using AVR. (Now, there,
I'm going to get called scum again!)

But I would love to try out your circuit!

Dank je wel!

--Mark

 

[mj]

mosfet driverhttp://ixdev.ixys.com/DataSheet/99061.pdf

D from BC
myrealaddress(at)comic(dot)com
BC, Canada
Posted to usenet sci.electronics.design

14 amps, well now that ought to do it. :{)

Can someone explain to me why a MOSFET would be better here than a
bipolar? Seems I should be getting enough current through the LED with
my 2n2222 darlington arrangement, but my transistor design skills have
always been modest. And when I learned it, MOSFET drivers were still
exotic, believe it or not. We did only bipolars.

 

[mj]

how about some of those phillips 1,3 or 5 watt leds used for maglight
replacement with a high current driver.

http://www.luxeonstar.com/

Bob

Those are nice, but they're kind of expensive. I'm actually doing a
dozen of these LEDs. And the .5W ones are *really* bright at 100mA.
If I can't get the 0.5 W ones to work, I'm going to look into tiny
xenon strobes.

 

[Jan Panteltje]

A PIC for blinking an LED? Not even LED*s*, but *an* LED?

You are scum among engineers!

(Obligatory ;-) included, since *I hope* you chose a PIC for timing
reasons or something.)

Tim

Well, yes, timing reasons and low part count.
Actually I looked up the little board, only one transistor..
and the PIC 12Fsomething.. just put it away again.
Indeed I wanted to see for myself if I could get better brightness from a white LED
with short very high current pulses, then with just DC.
It is not better,
I wanted the pulses to be at about mains frequency, or at least
faster then when you would see flashes.
555 timer could do it, but I have a load of those 12Fsomething PICs, and
no 555s, so why not use the PIC?
In my view the times of analog MVBs is over, big capacitors are more expensive then chips (PICs),
and you need to keep them in stock.
Sure you can do the same with 2 or 3 transistors... But not with the same
stability.

I do not see what people have against PICs, programmable logic, you see
people here make huge complex circuits with 7400 series or even CD400 series,
or whatever, as solutions to some question.
Was it you who did a z80 setup to just display some 7 segment data while
a PIC<-- could do it faster cheaper and better?
Just a few lines of asm in a micro (I am not married to PIC either), will do away with all
but one of the chips usually.
So when you have the assembler, the programmer, the PIC, and the knowledge how to write the code,
why go out an buy something you no longer need?
Even for the simplest thing,

 

[Jan Panteltje]

Actually, it's the circuit to pump current through those LEDs that I
need. My code works already. It works, but the light is feeble. The
asm source wouldn't do me much good, since I'm using AVR. (Now, there,
I'm going to get called scum again!)

But I would love to try out your circuit!

Dank je wel!

--Mark

Hi mark, I looked it up, it is very simple, a BCsomething switches the LED from
12V via a resistor to ground,.
Probably not advanced enough for when you want high efficiency.
The PIC drives the transistor via a base resistor, it was only one transistor I see now.
The PIC gets 5V via a 7805...

Het beste!

 

[Jan Panteltje]

14 amps, well now that ought to do it. :{)

Can someone explain to me why a MOSFET would be better here than a
bipolar? Seems I should be getting enough current through the LED with
my 2n2222 darlington arrangement, but my transistor design skills have
always been modest. And when I learned it, MOSFET drivers were still
exotic, believe it or not. We did only bipolars.

MOSFETs (special ones with low gate threshold voltage) interface nicely to
micros without drawing much current from the output pin.
At least for the low speeds that you are using.
MOSFET input capacitance can be as big as a few nF, so at higher speeds
switching get compromised (bad rise and fall times), say above a few kHz
for a pin that can source / sink 10 mA.
C.U = i.t

 

[Rich Grise]

I'm looking for ideas on how to make an LED flash so brightly at a low
duty cycle that it's reasonably bright--maybe even close to what it
would be if it were on DC.

Must be finals week again.

Good Luck!
Rich

 

[Rich Grise]

Actually, it's the circuit to pump current through those LEDs that I
need. My code works already. It works, but the light is feeble. The
asm source wouldn't do me much good, since I'm using AVR. (Now, there,
I'm going to get called scum again!)

But I would love to try out your circuit!

Doesn't anyone teach you googlies about the other side of google? I
just now put "circuit to pump current through an LED", without quotes,
into google's web search window, and got "about 227,000" hits.

Do your own homework.

Good Luck!
Rich

 

[Jon Kirwan]

<snip>
Actually, I'm actually doing this for a dozen LEDs at once. A rotating
transparent disk has [A-Z0-9 ] printed just inside the circumference
around the circle. A motor spins the disk at 20-50 RPM (controlled by
the MCU with PWM).
<snip>

Did you mean to say RPM?? At as low as 20 RPM, you'd be talking about
taking 3 seconds to make one revolution.

Jon

 

[Jon Kirwan]

Must be finals week again.

Not yet. It's the beginning of May, for gosh sake. Finals week is
probably towards the beginning of June.

Jon

 

[mj]

Doesn't anyone teach you googlies about the other side of google? I
just now put "circuit to pump current through an LED", without quotes,
into google's web search window, and got "about 227,000" hits.

Do your own homework.

Good Luck!
Rich

I tried adding extra snark to the circuit, and the LED is no
brighter. As I expected. Turns out snark sheds no light at all.

I did find some good articles on LED overdriving, but nothing at so
low a duty cycle.

It's not homework, just a personal project I get to work on a couple
of times a week.

But hey, thanks anyway.

 

[mj]

<snip>
Actually, I'm actually doing this for a dozen LEDs at once. A rotating
transparent disk has [A-Z0-9 ] printed just inside the circumference
around the circle. A motor spins the disk at 20-50 RPM (controlled by
the MCU with PWM).
<snip>

Did you mean to say RPM??  At as low as 20 RPM, you'd be talking about
taking 3 seconds to make one revolution.

Jon

Whoops, no--rps.

 

[Jon Kirwan]

<snip>
To try to get a bright enough flash, I got some 0.5W white LEDs that
can take a max DC forward current of 150 mA, and have about a ~3.6
forward drop, producing an intensity of 130k mcd. (Not too clear on
the mcd part.) And dem suckers is bright when you're pumping even 100
mA through them. Like squint-to-look-at-it bright. They're in a
standard 5mm package (T-1 3/4, what is up with that package name?),
though it's sturdier than most you've probably seen. The LED looks
like it's been lifting weights, and the leads are shorter and fatter.

Anyway, I have two 2n2222's hooked up as a Darlington, with +5 Vcc,
driving the front Q's base with ~15mA (with a microcontroller pin).
There's NO current-limiting resistor on the back end, where the second
transistor's collector is attached to +5v, and the emitter goes
through the LED to ground.
<snip>

Some thoughts about the 2N2222. They probably can handle the pulses
of 1.5A for 200us and .4% duty (I am assuming, for now, that your
20-50 RPM is really 20 RPS = 200us/0.4%.) However, the Rc, Re, and Rb
plus the 26mV*(1+ln(1+I/Is)) are going to suggest something on the
order of 1.9 to 2.0 volts at the base. (I see Re=0.2, Rc=0.3, and
Rb=10 for one model I have, with a blind Is=1e-14, and I don't think
you can expect to get better than beta=30 here.)

Vbe = 26e-3*(1+ln(1+(1.5/30)/1e-14))+1.5*(.2+.3)+(10+.2)*(1.5/30)

Which reads out at about 2V.

Since this is a darlington, I'm assuming something like this:

|
about 2.7V | <--- about 2.2V
| ,--------+
v | |
R1 |/c Q2 |
ON----/\/\-----| 2N2222 |
|>e |
| |/c Q1
about 2V --> '------| 2N2222
|>e
|
|
gnd

That's Q1, I'm talking about. The Vbe of Q2 is about 0.7V and, if Q2
is considered saturated at about Vce=0.2V, I'd expect to see a Vce on
Q1 of 2.2V or so. [Note that Q2's base is about 0.5V above Q2's
collector, so that is going to conduct a little (.2V/60mV, about 3
orders, or maybe 0.1% of the base current -- nothing to get excited
about.)]

With a 5V source above, that leaves about 2.8V for your LED. Not the
3.6V you were talking about at 150mA. And you want 1500mA, not 150mA!
So your circuit probably won't get there.

Here's the same circuit with 150mA as the estimate:

|
about 1.45V | <--- about 1V
| ,--------+
v | |
R1 |/c Q2 |
ON----/\/\-----| 2N2222 |
|>e |
| |/c Q1
about 0.85V --> '------| 2N2222
|>e
|
|
gnd

This provides about 4V compliance (not sure if you are using a
resistor on the collector leg) for your LED. So at 150mA, it probably
has just about enough for your 3.6V requirement.

.....

All this suggests to me that you aren't going to see a lot more than
150mA. I don't have the LED model, but since it is white the LED is
blue with some phosphor used for the white appearances. Guessing a
simple model has me taking about 2.8V as the minimum on-voltage (which
I get from (700nm/400nm)*1.6V, extrapolating from a red led) and a
model where R=(3.6-2.8)/150mA or in the area of 5 ohms. Actually,
that seems too high to me, so that 2.8V estimate is probably wrong.
Let's assume it is closer to about 2 ohms or so (3.3V minimum.) At
1.5A, that's still 3V all by itself. Adding that to the 3.3V figure
gives you well more than 5V. So again, a problem even assuming your
darlington arrangement could support a Q1 Vce of 0V, which it cannot.

Your darlington arrangement creates one problem. Q1's Vce will be on
the order of 2V at those high currents and you don't have that kind of
headroom to spare. Your LED itself creates another. It's not likely
to allow 1.5A with only 5V of drive, under any circumstance. These
are guesses, admittedly. But what this seems to say is that you
probably need another voltage rail, at a minimum, if you want to get
up to 1.5A on your LED.

If you can live with less than that but still want a lot more than
150mA, the perhaps your LED will allow ... hmm ... say (4.4V-3.3V)/2
.... or about 500mA, let's call it, assuming you can limit your BJT
switch to a Vce of 0.6V or less. And no more than (4.8V-3.3V)/2 or
about 700mA. I think that's the best you can hope for. But I'd plan
no more than 500mA, after some testing first. In this case, you can
keep your rail, maybe, but you have to lose the darlington.

Jon

 

[mj]

Well, yes, timing reasons and low part count.
Actually I looked up the little board, only one transistor..
and the PIC 12Fsomething.. just put it away again.
Indeed I wanted to see for myself if I could get better brightness from awhite LED
with short very high current pulses, then with just DC.
It is not better,
I wanted the pulses to be at about mains frequency, or at least
faster then when you would see flashes.
555 timer could do it, but I have a load of those 12Fsomething PICs, and
no 555s, so why not use the PIC?
In my view the times of analog MVBs is over, big capacitors are more expensive then chips (PICs),
and you need to keep them in stock.
Sure you can do the same with 2 or 3 transistors... But not with the same
stability.

I do not see what people have against PICs, programmable logic, you see
people here make huge complex circuits with 7400 series or even CD400 series,
or whatever, as solutions to some question.
Was it you who did a z80 setup to just display some 7 segment data while
a PIC<-- could do it faster cheaper and better?
Just a few lines of asm in a micro (I am not married to PIC either), willdo away with all
but one of the chips usually.
So when you have the assembler, the programmer, the PIC, and the knowledge how to write the code,
why go out an buy something you no longer need?
Even for the simplest thing,

Geez, I thought I was the scum! I agree--for one-offs, an 8-pin DIP is
an 8-pin DIP. For production, of course, it's all about total cost.
Often dominated by board size rather than component cost.

'swhy I like I2C so much -- wiring up discrete chip select logic (not
to mention bus lines) is a PITA.

--mj

 

[mj]

<snip>
To try to get a bright enough flash, I got some 0.5W white LEDs that
can take a max DC forward current of 150 mA, and have about a ~3.6
forward drop, producing an intensity of 130k mcd. (Not too clear on
the mcd part.) And dem suckers is bright when you're pumping even 100
mA through them. Like squint-to-look-at-it bright. They're in a
standard 5mm package (T-1 3/4, what is up with that package name?),
though it's sturdier than most you've probably seen. The LED looks
like it's been lifting weights, and the leads are shorter and fatter.

Anyway, I have two 2n2222's hooked up as a Darlington, with +5 Vcc,
driving the front Q's base with ~15mA (with a microcontroller pin).
There's NO current-limiting resistor on the back end, where the second
transistor's collector is attached to +5v, and the emitter goes
through the LED to ground.
<snip>

Some thoughts about the 2N2222.  They probably can handle the pulses
of 1.5A for 200us and .4% duty (I am assuming, for now, that your
20-50 RPM is really 20 RPS = 200us/0.4%.)  However, the Rc, Re, and Rb
plus the 26mV*(1+ln(1+I/Is)) are going to suggest something on the
order of 1.9 to 2.0 volts at the base.  (I see Re=0.2, Rc=0.3, and
Rb=10 for one model I have, with a blind Is=1e-14, and I don't think
you can expect to get better than beta=30 here.)

  Vbe = 26e-3*(1+ln(1+(1.5/30)/1e-14))+1.5*(.2+.3)+(10+.2)*(1.5/30)

Which reads out at about 2V.

Since this is a darlington, I'm assuming something like this:

                             |
      about 2.7V             | <--- about 2.2V
              |     ,--------+
              v     |        |
          R1      |/c Q2     |
   ON----/\/\-----|   2N2222 |
                  |>e        |
                    |      |/c Q1
       about 2V --> '------|   2N2222
                           |>e
                             |
                             |
                            gnd

That's Q1, I'm talking about.  The Vbe of Q2 is about 0.7V and, if Q2
is considered saturated at about Vce=0.2V, I'd expect to see a Vce on
Q1 of 2.2V or so.  [Note that Q2's base is about 0.5V above Q2's
collector, so that is going to conduct a little (.2V/60mV, about 3
orders, or maybe 0.1% of the base current -- nothing to get excited
about.)]

With a 5V source above, that leaves about 2.8V for your LED.  Not the
3.6V you were talking about at 150mA.  And you want 1500mA, not 150mA!
So your circuit probably won't get there.

Here's the same circuit with 150mA as the estimate:

                             |
      about 1.45V            | <--- about 1V
              |     ,--------+
              v     |        |
          R1      |/c Q2     |
   ON----/\/\-----|   2N2222 |
                  |>e        |
                    |      |/c Q1
    about 0.85V --> '------|   2N2222
                           |>e
                             |
                             |
                            gnd

This provides about 4V compliance (not sure if you are using a
resistor on the collector leg) for your LED.  So at 150mA, it probably
has just about enough for your 3.6V requirement.

....

All this suggests to me that you aren't going to see a lot more than
150mA.  I don't have the LED model, but since it is white the LED is
blue with some phosphor used for the white appearances.  Guessing a
simple model has me taking about 2.8V as the minimum on-voltage (which
I get from (700nm/400nm)*1.6V, extrapolating from a red led) and a
model where R=(3.6-2.8)/150mA or in the area of 5 ohms.  Actually,
that seems too high to me, so that 2.8V estimate is probably wrong.
Let's assume it is closer to about 2 ohms or so (3.3V minimum.)  At
1.5A, that's still 3V all by itself.  Adding that to the 3.3V figure
gives you well more than 5V.  So again, a problem even assuming your
darlington arrangement could support a Q1 Vce of 0V, which it cannot.

Your darlington arrangement creates one problem.  Q1's Vce will be on
the order of 2V at those high currents and you don't have that kind of
headroom to spare.  Your LED itself creates another.  It's not likely
to allow 1.5A with only 5V of drive, under any circumstance.  These
are guesses, admittedly.  But what this seems to say is that you
probably need another voltage rail, at a minimum, if you want to get
up to 1.5A on your LED.

If you can live with less than that but still want a lot more than
150mA, the perhaps your LED will allow ... hmm ... say (4.4V-3.3V)/2
... or about 500mA, let's call it, assuming you can limit your BJT
switch to a Vce of 0.6V or less.  And no more than (4.8V-3.3V)/2 or
about 700mA.  I think that's the best you can hope for.  But I'd plan
no more than 500mA, after some testing first.  In this case, you can
keep your rail, maybe, but you have to lose the darlington.

Jon

Thanks, Jon that's a very helpful analysis. I'll have to scope it out
in the next couple of days. And I'll consider a higher voltage rail
(that just means soldering that 7805 on the MCU board, where I'd left
it out before... oh, yeah, and probably decoupling caps if I'm
snapping off pulses like that...)

If 500-700 mA gives me the brightness I want, fine. I'm not looking to
laser etch anything. I just want LEDs that aren't pitfully dim. Your
discussion gives me hope that I can get there.

I actually don't have a current-limiting resistor on the collector
leg. When occasionally a software glitch leaves the LED on (I switch
it off quick), it's bright. Once it even started going a bit green,
but I shut it down just in time. (That color change is unmistakable.)
And it gets hot after just a few seconds.

As for the LED analysis, I'm only slowly figuring out how to model
LEDs in my head; I'll go through your discussion above in detail when
I have time.

Does anyone here have a favorite resource or book about circuit design
with LEDs? I've found a few good websites, but many of them are mostly
app notes for somebody's IC. I've had a great time with TLC5940s, but
they're overkill if all I want to do is flash.

Thanks again.

--Mark

 

[whit3rd]

I'm looking for ideas on how to make an LED flash so brightly at a low
duty cycle

Since the LED is only on 0.4% of the time, max, it still simply isn't
very bright. The strobe works--I can see the frozen image on the
spinning disk--but the light is simply anemic.

So, I'm wondering if anyone here knows how to design a circuit that
can dump an amp and a half through an LED for, say, 200 microseconds
at a time or less, at 20-50 Hz.

Firstly, I'd put a trickle through the LED at all times (maybe half a
milliamp)
so the storage capacitance doesn't have to be reloaded each flash.
Then, with a blocking diode, connect to a flyback switched inductor.
The current in the inductor builds, when it reaches Ipeak you
turn off the input, and the current continues to flow through the
only other connection, the LED with its blocking diode now forward
biased.

Taps on the inductor will allow impedance matching to both the
charging supply and the LED. It's just about the same kind of
circuit as an old auto ignition.

That's how the old flashlamp strobes worked; of course, the dynamic
range of a discharge lamp allows lower duty cycles, at really
large peak currents.