Digital Current Control for LED array

[theJackal]

On Wed, 21 Dec 2005 12:42:16 -0600, John Fields



Hey idiot ... you don't seem to even know how to read. If your mother
didnt send you to school then you might try sueing the long dead lady
but I'd advice you to
Go back to nursery school ...
I said I don't talk to idiots... so you're talking to yourself.

"Go easy on your drugs"

theJackal

 

[John Fields]

On Wed, 21 Dec 2005 12:42:16 -0600, John Fields



Hey idiot ... you don't seem to even know how to read. If your mother
didnt send you to school then you might try sueing the long dead lady
but I'd advice you to
Go back to nursery school ...
I said I don't talk to idiots... so you're talking to yourself.

---
didnt? sueing? advice you?

Seems like you could use a little help with the language yourself.

And as for the logic, you call me an idiot and in the same breath
you say you don't talk to idiots. And keep posting. And posting.
And posting.

You just can't keep from beating yourself up, can you, ya fuckin'
loser?

 

[theJackal]

Is this to correct for "shading" variances across the arrays?

How exactly do you intend to measure the actual light output of each LED?

Is this a one-off correction, or an ongoing process?

The only way the lamps can give the same light output is if they have the same current and
voltage across them assuming they are of the same type.

His idea of connecting them in parallel is wrong.
Check my earlier post on the matter.

"Go easy on the whisky"

theJackal

 

[John Fields]

The only way the lamps can give the same light output is if they have the same current and
voltage across them assuming they are of the same type.

---
You're wrong and, as usual, you don't even know why.
---

His idea of connecting them in parallel is wrong.
Check my earlier post on the matter.

---
Your earlier post was wrong and, as usual, you don't even know why.

Instead of posturing and pretending that you know something, you
really should get back to the basics and try to learn what it is you
need to know if you want to be in this game.

So far, everyone here has knocked you off as a rank amateur with
delusions of adequacy, but you seem to be too fucking stoopid to get
the message.

What's it gonna take to wise you up?

An effort to educate you seems inapproprite at this time since your
stance is hostile and and you seem to think that you're at the
center of the universe.

Oh well, no big loss...

 

[John_H]

theJackal wrote:

The only way the lamps can give the same light output is if they have the same current and
voltage across them assuming they are of the same type.

His idea of connecting them in parallel is wrong.
Check my earlier post on the matter.

"Go easy on the whisky"

theJackal

Two "lamps" with the same part number and same current will have
slightly different voltages because they're different. Close is good,
but it's not exact.

Two lamps with the same part number and same current selected to have
the "best match" voltage from a bin of parts will still have different
light output levels due to differences in phosphor coating (for white
LEDs) thermal conductivity of the package (or system) and other physical
characteristics of the LED materials, packaging, and environment.

The normalized luminance versus current curves for the individual
devices is similar in shape across production suggesting that a
first-order match of the LEDs by setting the current of each LED for a
fixed luminance will result in a "good" match (though not exact) as the
current is increased or decreased from the calibration point. The
differences of luminance at other proportional current levels and at
other thermal operating points are second-order mismatches.

By feeding each LED with a controllable current, the voltage will be
defined by the LED. These constant-current LEDs can be in series or
parallel requiring series or shunt regulation, respectively; it's just
easier to have series regulators in a parallel configuration. It also
makes the proportional control much simpler.

 

[theJackal]

theJackal wrote:


Two "lamps" with the same part number and same current will have
slightly different voltages because they're different. Close is good,
but it's not exact.

Note thats why I said above ... if they have the SAME current and voltage .

Two lamps with the same part number and same current selected to have
the "best match" voltage from a bin of parts will still have different
light output levels due to differences in phosphor coating (for white
LEDs) thermal conductivity of the package (or system) and other physical
characteristics of the LED materials, packaging, and environment.

In my first post in the thread I mentioned that.
I would include dopant density variations which affect the resistivities of
semiconductors.

The normalized luminance versus current curves for the individual
devices is similar in shape across production suggesting that a
first-order match of the LEDs by setting the current of each LED for a
fixed luminance will result in a "good" match (though not exact) as the
current is increased or decreased from the calibration point. The
differences of luminance at other proportional current levels and at
other thermal operating points are second-order mismatches.

Here you correctly say you set the currents of each LED for a fixed illuminance
the below you contradict yourself by saying a parallel or a series connection of LEDs is
fine. A parallel connection sets the Voltage of each LED as fixed and NOT the current.

The radiant power is a function of Photon emission. Photon emission is determined by hole
electron recombination currents in the quasi neutral n and p regions of the Light
emitting diodes.

By feeding each LED with a controllable current, the voltage will be
defined by the LED. These constant-current LEDs can be in series or
parallel requiring series or shunt regulation, respectively; it's just
easier to have series regulators in a parallel configuration. It also
makes the proportional control much simpler.

A series regulator in a parallel connection IS totally bad. Because you are changing not
only the voltage and current of 1 LED but of ALL the other LEDs . That doesn't happen if
its in a series connection.

"Go easy on the whisky"

theJackal

 

[John Fields]

Note thats why I said above ... if they have the SAME current and voltage .

---
No, thats _not_ what you said.

What you said was that two lamps with identical currents through
them and identical voltages across thenm will have the same light
output, and that's not right.
---

In my first post in the thread I mentioned that.
I would include dopant density variations which affect the resistivities of
semiconductors.

---
LOL, you fucking goofball. On the one hand you spout that
same-voltage same-current same-light-output bullshit and then turn
right around and say that the light output can vary because of
doping differences? You can't have it both ways.
---

Here you correctly say you set the currents of each LED for a fixed illuminance
the below you contradict yourself by saying a parallel or a series connection of LEDs is
fine. A parallel connection sets the Voltage of each LED as fixed and NOT the current.

---
No, it doesn't. In a parallel array the LEDs are connected in
parallel with the _supply_ and the current limiting resistors are
connected in _series_ with the LEDs and used to control the
luminance of the individual LEDs, like this:


+V>---+-----+-----+--//--+
| | | |
[R] [R] [R] [R]
| | | |
[LED] [LED] [LED] [LED]
| | | |
GND>--+-----+-----+--//--+
---

The radiant power is a function of Photon emission. Photon emission is determined by hole
electron recombination currents in the quasi neutral n and p regions of the Light
emitting diodes.

---
Puh-leeeze... Stop it already, trying to explain something you know
nothing about.
---

A series regulator in a parallel connection IS totally bad. Because you are changing not
only the voltage and current of 1 LED but of ALL the other LEDs . That doesn't happen if
its in a series connection.

---
Huh???

In a parallel array, as I showed above, each of the current limiting
resistors will control the current into only the LED to which it's
connected, while in a series string:

+V>---+
|
[R]
|
[LED]
|
[LED]
|
[LED]
|
[LED]
|

 

[John_H]

theJackal wrote:

Here you correctly say you set the currents of each LED for a fixed illuminance
the below you contradict yourself by saying a parallel or a series connection of LEDs is
fine. A parallel connection sets the Voltage of each LED as fixed and NOT the current.

The radiant power is a function of Photon emission. Photon emission is determined by hole
electron recombination currents in the quasi neutral n and p regions of the Light
emitting diodes.



A series regulator in a parallel connection IS totally bad. Because you are changing not
only the voltage and current of 1 LED but of ALL the other LEDs . That doesn't happen if
its in a series connection.

"Go easy on the whisky"

theJackal

I am suggesting 7 regulators for 7 LEDs. Of *course* connecting 7 LEDs
in parallel directly with one series regulator will not provide a
solution. Of *course* connecting 7 LEDs in series with one shunt
regulator will not work. If one "LED stage" is defines as an LED and a
regulator, 7 "LED stages" can be connected in parallel if the "LED stage
is built with a series regulator and 7 "LED stages" can be connected in
series if the "LED stage is built with a shunt regulator.

I don't contradict a thing. EACH LED is fed with a controllable
current. Separately. Not "all LEDs are fed with A controllable current."

 

[theJackal]

theJackal wrote:
<snipped>

I am suggesting 7 regulators for 7 LEDs. Of *course* connecting 7 LEDs
in parallel directly with one series regulator will not provide a
solution. Of *course* connecting 7 LEDs in series with one shunt
regulator will not work. If one "LED stage" is defines as an LED and a
regulator, 7 "LED stages" can be connected in parallel if the "LED stage
is built with a series regulator and 7 "LED stages" can be connected in
series if the "LED stage is built with a shunt regulator.

I don't contradict a thing. EACH LED is fed with a controllable
current. Separately. Not "all LEDs are fed with A controllable current."

If I get you right ... you are connecting a series resistor/regulator in each parallel
branch of the LED array.
Suppose you want to brighten a LED you'd decrease the resistance in that branch , but
that would increase the current flowing through it , which would then cause a decrease in
the currents flowing to the other LEDs ... so those would become dimmer . You'd then have
to work on the resistors in the other LED branches. Besides you'd have to take into
account the limiting current and the non linear voltage current characteristic of the LED.



theJackal






Any posts containing "John Fields" are automatically deleted by my system

 

[John Fields]

Any posts containing "John Fields" are automatically deleted by my system

 

[John_H]

If I get you right ... you are connecting a series resistor/regulator in
each parallel
branch of the LED array.
Suppose you want to brighten a LED you'd decrease the resistance in that
branch , but
that would increase the current flowing through it , which would then
cause a decrease in
the currents flowing to the other LEDs ... so those would become dimmer .
You'd then have
to work on the resistors in the other LED branches. Besides you'd have
to take into
account the limiting current and the non linear voltage current
characteristic of the LED.

<snip>

YES! You've got it now. The only major concept that you aren't seeing is
that a constant-current series regulator that uses feedback from the current
running through the LED *won't* be affected by the other LEDs current
changes as long as the supply is properly regulated. Constant-current loads
need a supply with a voltage that can meet the LED voltage drop plus
regulator "dropout voltage" for all the LEDs, supplying enough current to
accommodate all 7 LEDs at full current.

If you stop thinking "resistors" and start thinking "regulators" everything
can come together.

The need to ramp the intensity up and down with a single control - one
dimmer for 7 matched intensities - means that the regulators must track each
other as the brightness is increased and decreased. If the 7 regulators run
from 75% to 100% rated current to match brightnesses for a particular batch
of LEDs, half-power would servo those regulators to 37.5% to 50% of rated
current proportionally.

These high-current LEDs need a transistor (I'll use n-channel MOSFETS),
current sense (small value resistor for my implementation), an amplifier
with feedback to maintain a fixed current through the sense resistor through
the gate drive voltage, and a method to calibrate the individual LEDs and
present a "master" brightness. I may use potentiometers with tap to the
opamp and rails of ground and "master brigtness voltage" or use an 8-channel
multiplying DAC: change the reference voltage and the outputs scale.

The main supply just needs to supply a voltage that will have all the
regulators properly regulating (or all the op-amps out of saturation).

 

[John Fields]

If I get you right ... you are connecting a series resistor/regulator in each parallel
branch of the LED array.
Suppose you want to brighten a LED you'd decrease the resistance in that branch , but
that would increase the current flowing through it , which would then cause a decrease in
the currents flowing to the other LEDs ... so those would become dimmer . You'd then have
to work on the resistors in the other LED branches. Besides you'd have to take into
account the limiting current and the non linear voltage current characteristic of the LED.

---
You know what? You're about the stupidest sonofabitch I ever ran
across.

Here:

Ya take seven LEDs and hook them up in series with seven pots, like
this, and hook them all up in parallel with a _voltage_ source
(_not_ a current source) like this:


+V>----+--------+--------+--------+--------+--------+--------+
| | | | | | |
[POT]<-+ [POT]<-+ [POT]<-+ [POT]<-+ [POT]<-+ [POT]<-+ [POT]<-+
| | | | | | | | | | | | | |
+----+ +----+ +----+ +----+ +----+ +----+ +----+
| | | | | | |
[LED] [LED] [LED] [LED] [LED] [LED] [LED]
| | | | | | |
GND>---+--------+--------+--------+--------+--------+--------+

You take the maximum current you can put into the lamps (1 amp each,
according to the OP) and you figure out what the values of the
rheostats need to be based on your highest supply voltage and the
_nominal_ voltage drop of the lamps according to the spec sheet.

Let's say that you have a 10V supply which can put out 10 amps, and
that the Vf of the lamps is 3V. Then, with the supply cranked to
the max, the value of the rheostats needs to be:


Vs - Vf 10V - 3V
R = --------- = ---------- = 7 ohms
If 1A

And they'll each need to dissipate:


P = If² R = 1A² * 7R = 7 watts.


Now, since we know that all of the LEDs aren't going to be equally
bright what we'll need to determine is how wide the range of
brightnesses will be and compensate for that by using the rheostats
to dim all the brighter ones down to the level of the dimmest one.

Let's say that the brightest one can be twice as bright as the
dimmest one with the same current (1 amp) going through them both,
and that the relationship between luminance and current is linear
because the OP said so. That means to dim the bright one down
you'll have to limit the current through it to 500mA, which means
the rheostat's high resistance has to be:


Vs - Vf 10V - 3V
R = --------- = ---------- = 14 ohms
If 0.5A

and it'll have to dissipate


P = If² R = 0.5A² * 14R = 3.5 watts


So, under those conditions it looks like what's needed is about a 15
ohm 15 watt adjustable power resistor. 15 watts because that's
resistor's _entire_ winding is rated for, and since we're
dissipating 7 watts across half the winding we have to up the rating
by a factor of two to keep the resistor happy. Actually, 25 watts
would be a better choice for the money.

Finally, once the array is trimmed up and all the LEDs are set to
the same brightness, the output of the power supply (remeber the
power supply?) can be lowered to dim the entire array and the
brightness of all the LEDs should track.

So _that's_ how ya do it.
---

 

[Spehro Pefhany]

If you think about it, he's got to be lying. Otherwise, how did his
post get out? ;-)


Best regards,
Spehro Pefhany

 

[John Fields]

If you think about it, he's got to be lying. Otherwise, how did his
post get out? ;-)

 

[theJackal]

<snip>

YES! You've got it now. The only major concept that you aren't seeing is
that a constant-current series regulator that uses feedback from the current
running through the LED *won't* be affected by the other LEDs current
changes as long as the supply is properly regulated. Constant-current loads
need a supply with a voltage that can meet the LED voltage drop plus
regulator "dropout voltage" for all the LEDs, supplying enough current to
accommodate all 7 LEDs at full current.

If you stop thinking "resistors" and start thinking "regulators" everything
can come together.

The need to ramp the intensity up and down with a single control - one
dimmer for 7 matched intensities - means that the regulators must track each
other as the brightness is increased and decreased. If the 7 regulators run
from 75% to 100% rated current to match brightnesses for a particular batch
of LEDs, half-power would servo those regulators to 37.5% to 50% of rated
current proportionally.

These high-current LEDs need a transistor (I'll use n-channel MOSFETS),
current sense (small value resistor for my implementation), an amplifier
with feedback to maintain a fixed current through the sense resistor through
the gate drive voltage, and a method to calibrate the individual LEDs and
present a "master" brightness. I may use potentiometers with tap to the
opamp and rails of ground and "master brigtness voltage" or use an 8-channel
multiplying DAC: change the reference voltage and the outputs scale.

The main supply just needs to supply a voltage that will have all the
regulators properly regulating (or all the op-amps out of saturation).

I've just been able to have a rapid glimpse through your post , and it seems you've got a
completely different solution from what I had thought. As I have to run off for the
weekend away from here , will be seeing you on Monday or Tuesday.
Enjoy the holidays and don't overeat. <grin>

theJackal






Any posts containing "John Fields" are automatically deleted by my system

 

[Rich Grise]

If you think about it, he's got to be lying. Otherwise, how did his
post get out? ;-)

He doesn't know the difference between "delete" and "filter"?

Cheers!
Rich

 

[John_H]

Finally, once the array is trimmed up and all the LEDs are set to
the same brightness, the output of the power supply (remeber the
power supply?) can be lowered to dim the entire array and the
brightness of all the LEDs should track.

John Fields
Professional Circuit Designer

If the currents are supplied from a fixed voltage through resistors, when
you change the power supply voltage the brigtnesses will not track unless
you can guarantee an exact match of the LED voltages. I wouln't trust the
match to be sufficient for small voltage drops in the resistors. If one
chooses to use a large voltage drop across the resistors (low efficiency)
then the match is very good until the brightness is reduced significantly
and the LED voltages again become significant.

It may not be cost-effective to use potentiometers rated at the necessary
power levels to accommodate the large voltage drops. If the passive
approach were used, a combination of fixed resistors for a set point along
with more reasonable potentiometers might be warranted.

 

[John Fields]

If the currents are supplied from a fixed voltage through resistors, when
you change the power supply voltage the brigtnesses will not track unless
you can guarantee an exact match of the LED voltages. I wouln't trust the
match to be sufficient for small voltage drops in the resistors. If one
chooses to use a large voltage drop across the resistors (low efficiency)
then the match is very good until the brightness is reduced significantly
and the LED voltages again become significant.

It may not be cost-effective to use potentiometers rated at the necessary
power levels to accommodate the large voltage drops.

---
Yes, that's why toward the end of the article I wrote:

"So, under those conditions it looks like what's needed is about a
15 ohm 15 watt adjustable power resistor."

What I was thinking was along the lines of an adjustable vitreous
enamelled wirewound reesistor like a an Ohmite Type 210 or a
Huntington AVT, with a Huntington 25 ohm 50 watt unit available from
Digi-Key for $4.50 qty 1.

If the passive
approach were used, a combination of fixed resistors for a set point along
with more reasonable potentiometers might be warranted.

I agree, but I think that using a single adjustable fairly
high-power resistor instead of a rheostat and an assortment of fixed
resistors per LED would be even cheaper. Hard to beat $4.50 per
LED, no?

 

[John_H]

Yes, that's why toward the end of the article I wrote:

"So, under those conditions it looks like what's needed is about a
15 ohm 15 watt adjustable power resistor."

What I was thinking was along the lines of an adjustable vitreous
enamelled wirewound reesistor like a an Ohmite Type 210 or a
Huntington AVT, with a Huntington 25 ohm 50 watt unit available from
Digi-Key for $4.50 qty 1.


I agree, but I think that using a single adjustable fairly
high-power resistor instead of a rheostat and an assortment of fixed
resistors per LED would be even cheaper. Hard to beat $4.50 per
LED, no?

It's easy to get under $4.50 an LED and have efficiencies significantly
higher than the 30% you propose (10V supply, 3V load). The MOSFETs are
about $0.50 each, the 4-channel opamp is under $2 ($0.50 per channel), the
discrete resistors are negligible, and the cheap pots are down to $0.25
each. It isn't just a nest of wires but the supply can be made to
accommodate a very small dropout voltage for very high efficiencies. The
solution is elegant rather than brute force. I'd hate to throw away 50W of
power for 21W of light.

Most folks designing with the high-power Luxeon LumiLEDs are interested in
efficiency above convenience; otherwise incandescent lamps would be fine.

 

[John Fields]

It's easy to get under $4.50 an LED and have efficiencies significantly
higher than the 30% you propose (10V supply, 3V load). The MOSFETs are
about $0.50 each, the 4-channel opamp is under $2 ($0.50 per channel), the
discrete resistors are negligible, and the cheap pots are down to $0.25
each. It isn't just a nest of wires but the supply can be made to
accommodate a very small dropout voltage for very high efficiencies. The
solution is elegant rather than brute force. I'd hate to throw away 50W of
power for 21W of light.

Most folks designing with the high-power Luxeon LumiLEDs are interested in
efficiency above convenience; otherwise incandescent lamps would be fine.