Maker Pro
Maker Pro

PWM controlled current sink for LED backlight.

S

Sean Peters

Jan 1, 1970
0
Hello,

I have a monitor with a LED backlight that uses PWM to directly switch
on/off the LEDs to vary the brightness.

I can't stand the flicker this produces and want to modify the LED
driver to keep the LEDs on all the time, setting the brightness by
limiting the drive current, depending on a smoothed version of the PWM
signal.

I've kind of managed this already, but can't get the full output
brightness range because i am using a LM324 op amp with a single 5v
supply and it can't pull it's output down below 0.6V while sinking
current. (Only existing easy to get at supply is 5V). Here's some
details of what i've found / tried so far:

The monitor is a ViewSonic VLED221wm, which contains a CHI MEI
A220Z2-H01 panel and Asian Power devices FM-60L12-AAB power supply.

The LEDs are driven by 3 MBI1816GT drivers (separate R, G, B). These
drivers have an active low output enable (!OE) pin and a pin where you
are supposed to put a resistor (Rext) to ground to control the current
output. (R,G=2K9, B=3K6 ohms). The data sheet doesn't have any
internal details or suggest anything other than just using a resistor,
it gives the current for each LED as 59*1.24/Rext and there is 1.24V
on this pin and by varying the voltage of the "grounded" end of the
resistors (driven from an op-amp) i can change the output led
brightness. I'm pretty sure the output current is just 59xRref
current.

There is a 7way connector coming from some chip i can't get to on the
side of the lcd panel. The pins seem to be a i2c bus, power, gnd and
R, G, B PWM signals. The PWM signals go via transistors to the !OE
pins. This chip must have colour sensors as it adjusts the duty cycle
of the PWM to try to get the backlight to be white as well as at the
correct overall brightness. (found by manually changing the value of
Rref for one colour and watching it try to compensate with the others).

So, what i've done so far is: tie low !OE inputs, for each R,G,B, take
the !PWM signal that used to go to !OE, pullup with 1K, low pass
filter it with R=100K, C=1uF, divide this by 4 with a 1M pot, buffer
with op-amp follower and connect this to what used to be the ground
end of the Rref resistors. Here's a diagram: (need fixed-width font)

------+-------------------------------------------------- 5V
|
+++
| |Rpullup(1K)
| |
+++ (want 0-1.24V)
| +----+ 0.2-4.8V +--------+
------+---+ +----+-------+ | |
!PWM +----+ | | | |`. | Rref=2K9/3K6
R=100k | +++ +-+- `. | +----+
--+-- | | | >-+--+ +----- Rref
--+-- | |<-----++ ,' +----+
C=1uF| | | |,'LM324
| +++1M
| |
--------------------+-------+---------------------------- GND

I was very pleased to find that this actually works, time constant is
probably a bit too big as the backlight colours sometimes oscillate a
bit after turning on. The averaged !PWM varies from 0.2V (led
brightest) to 4.8V (led dimmest) divided down and buffered gives 1.2V
for dimmest and should give .05V for brightest. Which would give Rref
current from (1.24-1.2)/2900 to (1.2-0.05)/2900.

Only problem is i only get about half the range of brightness i should
as the LM324 can't take it's output below 0.63V as it is sinking
current (0.43mA max). Max Rref current is (1.24-0.63)/2900.

So, can you help? Is there an op-amp i can use instead that will
drive it's output down to 0V while sinking current with a single 5V
supply? Is there a better way of doing this? Any suggestions would
be greatly appreciated.

Also, i'm wondering, is it normal to drive LED's directly with a PWM
signal like this, or have i just been unlucky? I've got a weird
problem with my vision which makes me extremely sensitive to flicker i
can't use crt screens at all or CCFL backlight for more than a couple
of minutes. I just assumed LEDs would be driven by nice friendly DC.

Sean
 
S

Sjouke Burry

Jan 1, 1970
0
Sean said:
Hello,

I have a monitor with a LED backlight that uses PWM to directly switch
on/off the LEDs to vary the brightness.

I can't stand the flicker this produces and want to modify the LED
driver to keep the LEDs on all the time, setting the brightness by
limiting the drive current, depending on a smoothed version of the PWM
signal.

I've kind of managed this already, but can't get the full output
brightness range because i am using a LM324 op amp with a single 5v
supply and it can't pull it's output down below 0.6V while sinking
current. (Only existing easy to get at supply is 5V). Here's some
details of what i've found / tried so far:

The monitor is a ViewSonic VLED221wm, which contains a CHI MEI
A220Z2-H01 panel and Asian Power devices FM-60L12-AAB power supply.

The LEDs are driven by 3 MBI1816GT drivers (separate R, G, B). These
drivers have an active low output enable (!OE) pin and a pin where you
are supposed to put a resistor (Rext) to ground to control the current
output. (R,G=2K9, B=3K6 ohms). The data sheet doesn't have any
internal details or suggest anything other than just using a resistor,
it gives the current for each LED as 59*1.24/Rext and there is 1.24V
on this pin and by varying the voltage of the "grounded" end of the
resistors (driven from an op-amp) i can change the output led
brightness. I'm pretty sure the output current is just 59xRref
current.

There is a 7way connector coming from some chip i can't get to on the
side of the lcd panel. The pins seem to be a i2c bus, power, gnd and
R, G, B PWM signals. The PWM signals go via transistors to the !OE
pins. This chip must have colour sensors as it adjusts the duty cycle
of the PWM to try to get the backlight to be white as well as at the
correct overall brightness. (found by manually changing the value of
Rref for one colour and watching it try to compensate with the others).

So, what i've done so far is: tie low !OE inputs, for each R,G,B, take
the !PWM signal that used to go to !OE, pullup with 1K, low pass
filter it with R=100K, C=1uF, divide this by 4 with a 1M pot, buffer
with op-amp follower and connect this to what used to be the ground
end of the Rref resistors. Here's a diagram: (need fixed-width font)

------+-------------------------------------------------- 5V
|
+++
| |Rpullup(1K)
| |
+++ (want 0-1.24V)
| +----+ 0.2-4.8V +--------+
------+---+ +----+-------+ | |
!PWM +----+ | | | |`. | Rref=2K9/3K6
R=100k | +++ +-+- `. | +----+
--+-- | | | >-+--+ +----- Rref
--+-- | |<-----++ ,' +----+
C=1uF| | | |,'LM324
| +++1M
| |
--------------------+-------+---------------------------- GND

I was very pleased to find that this actually works, time constant is
probably a bit too big as the backlight colours sometimes oscillate a
bit after turning on. The averaged !PWM varies from 0.2V (led
brightest) to 4.8V (led dimmest) divided down and buffered gives 1.2V
for dimmest and should give .05V for brightest. Which would give Rref
current from (1.24-1.2)/2900 to (1.2-0.05)/2900.

Only problem is i only get about half the range of brightness i should
as the LM324 can't take it's output below 0.63V as it is sinking
current (0.43mA max). Max Rref current is (1.24-0.63)/2900.

So, can you help? Is there an op-amp i can use instead that will
drive it's output down to 0V while sinking current with a single 5V
supply? Is there a better way of doing this? Any suggestions would
be greatly appreciated.

Also, i'm wondering, is it normal to drive LED's directly with a PWM
signal like this, or have i just been unlucky? I've got a weird
problem with my vision which makes me extremely sensitive to flicker i
can't use crt screens at all or CCFL backlight for more than a couple
of minutes. I just assumed LEDs would be driven by nice friendly DC.

Sean
Why not (only if the led has a series resistor, or a current source),
put a cap of about 100 microf accross it?
Or is that to simple :)
 
P

pimpom

Jan 1, 1970
0
Sean Peters said:
Hello,

I have a monitor with a LED backlight that uses PWM to directly
switch
on/off the LEDs to vary the brightness.

I can't stand the flicker this produces and want to modify the
LED
driver to keep the LEDs on all the time, setting the brightness
by
limiting the drive current, depending on a smoothed version of
the PWM
signal.

I've kind of managed this already, but can't get the full
output
brightness range because i am using a LM324 op amp with a
single 5v
supply and it can't pull it's output down below 0.6V while
sinking
current. (Only existing easy to get at supply is 5V). Here's
some
details of what i've found / tried so far:

The monitor is a ViewSonic VLED221wm, which contains a CHI MEI
A220Z2-H01 panel and Asian Power devices FM-60L12-AAB power
supply.

The LEDs are driven by 3 MBI1816GT drivers (separate R, G, B).
These
drivers have an active low output enable (!OE) pin and a pin
where you
are supposed to put a resistor (Rext) to ground to control the
current
output. (R,G=2K9, B=3K6 ohms). The data sheet doesn't have
any
internal details or suggest anything other than just using a
resistor,
it gives the current for each LED as 59*1.24/Rext and there is
1.24V
on this pin and by varying the voltage of the "grounded" end of
the
resistors (driven from an op-amp) i can change the output led
brightness. I'm pretty sure the output current is just 59xRref
current.

There is a 7way connector coming from some chip i can't get to
on the
side of the lcd panel. The pins seem to be a i2c bus, power,
gnd and
R, G, B PWM signals. The PWM signals go via transistors to the
!OE
pins. This chip must have colour sensors as it adjusts the
duty cycle
of the PWM to try to get the backlight to be white as well as
at the
correct overall brightness. (found by manually changing the
value of
Rref for one colour and watching it try to compensate with the
others).

So, what i've done so far is: tie low !OE inputs, for each
R,G,B, take
the !PWM signal that used to go to !OE, pullup with 1K, low
pass
filter it with R=100K, C=1uF, divide this by 4 with a 1M pot,
buffer
with op-amp follower and connect this to what used to be the
ground
end of the Rref resistors. Here's a diagram: (need fixed-width
font)

------+-------------------------------------------------- 5V
|
+++
| |Rpullup(1K)
| |
+++ (want 0-1.24V)
| +----+ 0.2-4.8V +--------+
------+---+ +----+-------+ | |
!PWM +----+ | | | |`. | Rref=2K9/3K6
R=100k | +++ +-+- `. | +----+
--+-- | | | >-+--+ +----- Rref
--+-- | |<-----++ ,' +----+
C=1uF| | | |,'LM324
| +++1M
| |
--------------------+-------+---------------------------- GND

I was very pleased to find that this actually works, time
constant is
probably a bit too big as the backlight colours sometimes
oscillate a
bit after turning on. The averaged !PWM varies from 0.2V (led
brightest) to 4.8V (led dimmest) divided down and buffered
gives 1.2V
for dimmest and should give .05V for brightest. Which would
give Rref
current from (1.24-1.2)/2900 to (1.2-0.05)/2900.

Only problem is i only get about half the range of brightness i
should
as the LM324 can't take it's output below 0.63V as it is
sinking
current (0.43mA max). Max Rref current is (1.24-0.63)/2900.

So, can you help? Is there an op-amp i can use instead that
will
drive it's output down to 0V while sinking current with a
single 5V
supply? Is there a better way of doing this? Any suggestions
would
be greatly appreciated.

Also, i'm wondering, is it normal to drive LED's directly with
a PWM
signal like this, or have i just been unlucky? I've got a
weird
problem with my vision which makes me extremely sensitive to
flicker i
can't use crt screens at all or CCFL backlight for more than a
couple
of minutes. I just assumed LEDs would be driven by nice
friendly DC.

Sean

There are quite a few opamps with near rail-to-rail output
swings. NatSem has several. Then there are the venerable
CA3130/3140. These operate at 5V and have outputs of less than
100mV sinking the current levels you mentioned.
 
N

Nicholas Sherlock

Jan 1, 1970
0
Sean said:
I have a monitor with a LED backlight that uses PWM to directly switch
on/off the LEDs to vary the brightness.

I can't stand the flicker this produces and want to modify the LED
driver to keep the LEDs on all the time, setting the brightness by
limiting the drive current, depending on a smoothed version of the PWM
signal.

The monitor is a ViewSonic VLED221wm, which contains a CHI MEI
A220Z2-H01 panel and Asian Power devices FM-60L12-AAB power supply.

According to the user manual, you can disable this feature by holding
down the "Up triangle button" for 3 seconds.

Cheers,
Nicholas Sherlock
 
S

Sean Peters

Jan 1, 1970
0
Sjouke Burry said:
Why not (only if the led has a series resistor, or a current source),
put a cap of about 100 microf accross it?
Or is that to simple :)

Thanks for the suggestion, you're right, it would work, but the 3 LED
drivers each have 16 outputs. I don't think there'd be space for the
caps even if my soldering skills were up to it.

Sean
 
S

Sean Peters

Jan 1, 1970
0
pimpom said:
There are quite a few opamps with near rail-to-rail output
swings. NatSem has several. Then there are the venerable
CA3130/3140. These operate at 5V and have outputs of less than
100mV sinking the current levels you mentioned.

Thanks for the pointers, i went with the CA3140 in the end as it was
easiest to find.

Sean
 
S

Sean Peters

Jan 1, 1970
0
Nicholas Sherlock said:
According to the user manual, you can disable this feature by holding
down the "Up triangle button" for 3 seconds.

Cheers,
Nicholas Sherlock

Yes, That stops the backlight adapting to the brightness of what's on
the screen. What i'm trying to stop is the high frequency flicker
from the PWM modulated backlight (Most people wont even notice this).

Sean
 
Sean, you are super man.
what's the frequency of the flicker that you can sense?
PWM dimming method is widely used nowadays, that's bad news to you.
 
S

Sean Peters

Jan 1, 1970
0
Sean, you are super man.
what's the frequency of the flicker that you can sense?
PWM dimming method is widely used nowadays, that's bad news to you.

The frequency of the PWM in this monitor is 6.2kHz.

This LED backlit monitor was probably the best i'd used yet, but the
flicker was still a big problem for me, now i've filtered the PWM,
it's brilliant!
PWM dimming method is widely used nowadays, that's bad news to you.

That's a shame, i thought that now LED's were beginning to take over
from other light sources i'd be better off.

In case you're wondering: I've got a problem where i end up with lots
of noise in my vision. The intensity of the noise is dependant on the
speed of changes in what i'm looking at. The noise can build up to a
level where i can't see. Flicker causes me a huge problem. Low
frequency strip lights are the absolute worst, and old crt tv's.
higher frequency tv's and monitors are slightly better, then compact
fluorescent, ccfl, high frequency strip lights ... etc. Incandecent
bulbs seem to be the only lights i don't have a problem with.

Sean
 
M

Matthew

Jan 1, 1970
0
The frequency of the PWM in this monitor is 6.2kHz.

This LED backlit monitor was probably the best i'd used yet, but the
flicker was still a big problem for me, now i've filtered the PWM,
it's brilliant!


That's a shame, i thought that now LED's were beginning to take over
from other light sources i'd be better off.

In case you're wondering: I've got a problem where i end up with lots
of noise in my vision. The intensity of the noise is dependant on the
speed of changes in what i'm looking at. The noise can build up to a
level where i can't see. Flicker causes me a huge problem. Low
frequency strip lights are the absolute worst, and old crt tv's.
higher frequency tv's and monitors are slightly better, then compact
fluorescent, ccfl, high frequency strip lights ... etc. Incandecent
bulbs seem to be the only lights i don't have a problem with.

Sean

I am working for a ccfl driver producer, some of the ccfl monitors in
the present market use analog dimming. i don't know if you still have
problem with this kind of monitor. or maybe led backlight is your best
choice.

By the way, can you watch film in a cinema? I think a film has a
really low frequency flicker.
 
S

Sean Peters

Jan 1, 1970
0
I am working for a ccfl driver producer, some of the ccfl monitors in
the present market use analog dimming. i don't know if you still have
problem with this kind of monitor. or maybe led backlight is your best
choice.

I've not found a flourescent light of any kind that is not a problem
to me. I've been assuming that it's physically not possible to drive
a flourescent light in a way that prevents it from flickering. Is it
possible? Do you have any spectrum analysis results of fluorescent
light flicker?
By the way, can you watch film in a cinema? I think a film has a
really low frequency flicker.

Actually i've only been to the cinema a couple of times since i've had
this problem (about 5 years ago, it started after i got Glandular
Fever). But from what i can remember, the flicker wasn't really a big
problem. Not as bad for me as a monitor.

Sean
 
B

Bob Myers

Jan 1, 1970
0
I've not found a flourescent light of any kind that is not a problem
to me. I've been assuming that it's physically not possible to drive
a flourescent light in a way that prevents it from flickering. Is it
possible? Do you have any spectrum analysis results of fluorescent
light flicker?

Any such light source will have some level of flicker
component in its output; the question is whether or not
the flicker will be perceptible. I seriously question whether
it is actually flicker that gives you your trouble with
fluorescent lights, at least in the case of LCD backlighting.
Actually i've only been to the cinema a couple of times since i've had
this problem (about 5 years ago, it started after i got Glandular
Fever). But from what i can remember, the flicker wasn't really a big
problem. Not as bad for me as a monitor.

Which again makes me think that it's not flicker that's
causing you a problem with monitors. Movies are
FAR worse in terms of flicker than an LCD.

Bob M.
 
J

Jasen Betts

Jan 1, 1970
0
I've not found a flourescent light of any kind that is not a problem
to me. I've been assuming that it's physically not possible to drive
a flourescent light in a way that prevents it from flickering. Is it
possible? Do you have any spectrum analysis results of fluorescent
light flicker?

it's possible - you can drive them with DC, but then you get bright
and dark bands, it looks kind of neat as the band move as you adjust
the voltage, but I don't think it'svery good for the tube.

using a higher AC frequency should reduce the visible flicker

48 or 72 hz depends how many lobes on the shutter.

Bye.
Jasen
 
Top