Searching \ for '[PIC] PWM over a Luxeon LED' in subject line. ()
Make payments with PayPal - it's fast, free and secure! Help us get a faster server
FAQ page: www.piclist.com/techref/microchip/ios.htm?key=pwm
Search entire site for: 'PWM over a Luxeon LED'.

Exact match. Not showing close matches.
PICList Thread
'[PIC] PWM over a Luxeon LED'
2007\06\06@144738 by Ariel Rocholl

picon face
Hi there,

I am controlling a number of Luxeon LED with PWM using PIC12F family
capabilities.

I've been using a simple LM317 constant current configuration and now wanted
to move to a DC-DC circuit to reduce power loss.

Now the 1st question: In my current LM317 setup, I am using a FET to switch
the whole constant current circuit on/off as per PWM duty cycle, and I am
not sure that is the right way. I mean I have two options, right? I am
currently using option a) below

a) PWM switches on/off the constant current circuit, or
b) PWM switches on/off the LOAD, having the constant current circuit
connected all the time.

It was somehow easier to guarantee the LM317 is not having any option to
work on the off side, but I wonder if a LM317 is designed to be switched at
10KHZ in its input - probably not. So maybe is a more adequate solution to
move to option b).

Now the same question but for a DC-DC circuit (step down): I would think
using option b) is again more natural, but I may need to protect the FET
with a diode.

Thanks in advance
--
Ariel Rocholl
Madrid, Spain

2007\06\06@151127 by Harold Hallikainen

face
flavicon
face
Since you're using an LM317 as the current source, I'll assume your supply
voltage is higher than the LED voltage. So, how about this for a circuit?

+supply-->inductor-->LED-->FET Drain

FET Source-->small current sense R-->ground

FET Drain-->kDiodeA--> +supply

Turn on the FET when the FET source current is below the LED current you
want. Turn off the FET when the FET source current is above the current
you want.

Harold

{Quote hidden}

> -

2007\06\07@045423 by Ariel Rocholl

flavicon
face
I think you meant diode opposite right? FET Drain-->ADiodeK--> +supply

Nevertheless, your proposal is a good simple switching regulator, but I do
have already a buck DC-DC converter that theoretically can get about 85%
efficiency, so I just want to use it but connecting the FET driver properly.

Would you recommend configuration a) or b)?


2007/6/6, Harold Hallikainen <spam_OUTharoldTakeThisOuTspamhallikainen.org>:
{Quote hidden}

2007\06\07@054639 by Chris McSweeny

picon face
I'd recommend configuration c) - using a buck driver which has an external
current control, and drive that by low pass filtering the PWM output of the
PIC to give you a voltage to feed in.

On 6/7/07, Ariel Rocholl <.....forosKILLspamspam@spam@arocholl.com> wrote:
{Quote hidden}

>

2007\06\07@064927 by Ariel Rocholl

picon face
Chris, can you elaborate a bit on the advantage of c) configuration? Is it
to avoid any switching of the driver (by using the low pass filter) worth
the component count increment?


2007/6/7, Chris McSweeny <.....cpmcsweenyKILLspamspam.....gmail.com>:
{Quote hidden}

> >

2007\06\07@065004 by Ariel Rocholl

flavicon
face
Chris, can you elaborate a bit on the advantage of c) configuration? Is it
to avoid any switching of the driver (by using the low pass filter) worth
the component count increment?

2007/6/7, Chris McSweeny <@spam@cpmcsweenyKILLspamspamgmail.com>:
{Quote hidden}

> >

2007\06\07@071558 by Russell McMahon

face
flavicon
face
> I am controlling a number of Luxeon LED with PWM using PIC12F family
> capabilities.
>
> I've been using a simple LM317 constant current configuration and
> now wanted
> to move to a DC-DC circuit to reduce power loss.

Whatever arrangement you use should always run the Luxeon at full
current when it is running, with the duty cycle altering to control
brightness. This is because the LED colour and brightness per mA vary
as current varies and they are optimised and specified at full
current.

If you use PWM to produce a lower than rated current on average, and
then filter this so that the LED sees DC, rather than seeing PWM at
full current / no current then you get inferior results.



       Russell


2007\06\07@072211 by Chris McSweeny

picon face
Two reasons - it does avoid switching the driver on and off at high
frequency which won't actually be that good for a switch mode driver - note
that with a switch mode current driver you do really want to this this on
the supply side, so as not to leave it trying to drive current into an open
circuit. Also it is more efficient to drive LEDs at lower current than at
full current with PWM switching - typically when using current dimming you
get 60% of the light for only 50% of the power, whereas using PWM dimming
you only get 50% of the light for 50% of the power, so a 20% efficiency
improvement.

BTW the low pass filter required to convert a PWM output to a voltage is
only a single resistor and capacitor (or two resistors and one capacitor if
you also want to divide the voltage down to a lower level), so hardly vast
amounts of components.

On 6/7/07, Ariel Rocholl <spamBeGonearochollspamBeGonespamgmail.com> wrote:
{Quote hidden}

> > >

2007\06\07@072955 by Chris McSweeny

picon face
Inferior how? Yes the colour does change with varying current, but it's
rather strange to suggest it's optimised at full current - that's simply the
point it is specced at. You can in fact typically get a different binned LED
which will have the same colour at 350mA as your original one does at 700mA.
In any case, the effect is actually pretty minimal and something you can
pretty much ignore unless you're using the LEDs to illuminate artwork. The
other question of course is what exactly is "full current", given you often
get specs for more than one drive current.

Even more bizarre is your mention of brightness per mA varying with
different drive currents and this being optimised at full drive current. In
fact LED efficiency (since this is what you're referring to) improves with
decreasing drive currents, so by "optimised" you can only possibly mean
"minimised" :-)

On 6/7/07, Russell McMahon <EraseMEapptechspamparadise.net.nz> wrote:
{Quote hidden}

> -

2007\06\07@083802 by Russell McMahon

face
flavicon
face
I thought about digging up some references, but I was sure someone
would have more fun trashing what I said if I didn't (and it takes
time and ...) :-)

Maybe I should have said something like - "Philips, who manufacture
Luxeon LEDs say that ..."

{Quote hidden}

Your comments about Lumens per mA are correct or very correct in some
instances and wrong or very wrong in others.
Light output per mA is typically of the form [Philps say]
(Iactual/Ispecified)^k where typically 1 <= k <= 2.
In many cases in the range of interest [Philips say] light output with
current tends to be reasonably linear around specified current but non
linear and smaller at lower currents. cf reference 2 below, expecially
pp 16-21 and figure 3-17. That said, I was looking at the
specification for some CREE 1 to 5W LEDs recently - direct competitors
for Luxeons in many applications - and their optimum Lumens per Watt
is indeed at 350 mA with a significantly lower figure at 700 mA.

Re PWMing versus linear drive - Below are Philips bizzarre
recomendations from a number of their Luxeon documents.
   If brightness matching between LEDs and retention of the
manufacturers binning across a significant range of brightnesses is
not a requirement them the following is less applicable.

The second document below is liable to be useful reading for anyone
wanting to use Luxeons and similar "in anger".

To give a real world example of where your assertion is correct
(albeit not by a vast amount) the Philips Luxeon 3 data sheet here
http://www.luxeon.com/pdfs/DS45.PDF page 12 figs 6 & 7 give normalised
luminous flux versus current and the curves do show a degradation of
flux per mA as mA increases.

Another factor which adds complexity is the degradation of luminous
output with die temperature. The resultant die temperature for PWM and
AM schemes depends not only on mean power but on mechanical factors. I
haven't looked into this aspect in enough detail to make useful
comment.


       Russell

_______________________________________________

1.    From http://www.luxeon.com/pdfs/AB27.PDF


AM

Amplitude Modulation (AM) varies the current that is driven
through an LED string.

AM changes the brightness of the LEDs by reducing the
current through the LEDs. That concept works well where
the LED light output is well correlated to the nominal LED
drive current via production flux testing. Lumileds bins 100%
of their LEDs for flux at the nominal LED drive current. Thus
there is a strong correlation between the brightness of the
LEDs and its drive current in the range of the nominal LED
current. However, when the drive current drops below 50%
of the nominal LED current, then the light output is not as
well correlated with the LED drive current anymore and thus
the brightness variations from LED to LED tend to increase.

Because of this it is not recommended to exceed an AM
dimming ratio of 2:1.

Pulse Width Modulation

The Pulse Width Modulation (PWM) method drives the LEDs
at their nominal current and varies the duty cycle to vary the
average current through the LED to match the desired
brightness. Because with PWM the LEDs are always driven
at their nominal current, there are no correlation issues with
the LED brightness and its drive current. This method
provides for a linear relationship between the pulse width,
the average LED current and the brightness of the LEDs.

Since LEDs have a turnon time of less than 100ns, the
dimming ratio for PWM driven LED systems is only limited by
the rise or fall time of the driver. Theoretically the maximum
dimming ratio is 1:35000, while in practice, a dimming ratio
of 1:100 is more commonly used.

The rise and fall time of a driver output is defined as the time
that is needed to increase the current from 10% of the peak
current to 90% and vice versa. See Figure 5.2. The dimming
is limited to the point where the pulse width becomes shorter
than the rise or fall time of the driver.

_____________


2.    http://www.luxeon.com/pdfs/AB20-3.pdf

For best matching, it is recommended that a
pulse width modulation (PWM) circuit be
designed to operate the signal lamp at
reduced luminous intensity. The Stop signal
might operate the LED array at a high DC
forward current. Then for the Tail signal, the
array would be operated at the same peak
forward current with a low duty cycle (ratio of
.on. time to .on. plus .off. time). This
approach provides light output matching under
both levels of luminous intensity. A
recommended PWM circuit is shown in the
section .Special Considerations for Dual
Luminous Intensity Operation."

2007\06\07@104740 by Chris McSweeny

picon face
On 6/7/07, Russell McMahon <RemoveMEapptechEraseMEspamEraseMEparadise.net.nz> wrote:
>
> That said, I was looking at the
> specification for some CREE 1 to 5W LEDs recently - direct competitors
> for Luxeons in many applications - and their optimum Lumens per Watt
> is indeed at 350 mA with a significantly lower figure at 700 mA.


Still not optimum at 350mA - that's simply the lower  current which they bin
the LEDs at. Efficiency is better yet for these LEDs at even lower drive
currents. If we're going to play references, then see the top figure on page
5 of http://www.cree.com/products/pdf/XLamp7090XR-E.pdf :-) Or better yet,
here's some real world testing
http://www.candlepowerforums.com/vb/showthread.php?t=152094 where as you can
see the luminous efficiency is increasing right down to the lowest current
measurement at 20mA, and shows the effect is actually pretty significant
(50% better efficiency at 350mA compared to 1A). I'm still not sure what
evidence you've given to prove the assertion that I'm wrong, let alone very
wrong in my comments about efficiency improving with lower drive levels -
plenty of evidence for my position.

Looking at those docs of yours, it's obvious there are cases where keeping
the colour the same is important, such as in displays. Meanwhile I'm
interested in the very specialist area of battery powered high intensity
lighting where efficiency is everything - of course my needs may not be the
same as Ariel's. I should point out though that whatever those docs of yours
suggest I notice no issues at all with running strings of 3 or 4 Cree LEDs
at currents between 60mA and 1A - if the brightness varies between the
different LEDs in the string I can't say it's something I've noticed given
I'm using the LEDs to illuminate rather than staring into them!

Chris

More... (looser matching)
- Last day of these posts
- In 2007 , 2008 only
- Today
- New search...