On Sun, 22 May 2011 18:33:13 -0700 (PDT), linnix <me@linnix.info-for.us>
wrote:

>On May 22, 6:14&#4294967295;pm, Rich Webb <bbew...@mapson.nozirev.ten> wrote:
>> On Sun, 22 May 2011 17:19:34 -0700, D Yuniskis
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> <not.going.to...@seen.com> wrote:
>> >On 5/22/2011 4:31 PM, linnix wrote:
>> >> On May 22, 4:22 pm, Rich Webb<bbew...@mapson.nozirev.ten> &#4294967295;wrote:
>> >>> On Sun, 22 May 2011 15:34:58 -0700 (PDT), Gene<profii...@gmail.com>
>> >>> wrote:
>>
>> >>>> On Sunday, May 22, 2011 4:58:42 PM UTC-4, D Yuniskis wrote:
>> >>>>> [grrr... how about learning to break your lines at something like
>> >>>>> 70 characters?? &#4294967295;I've taken the liberty to *chop* off everything that
>> >>>>> extends beyond the width of my window... no fun, eh? &#4294967295;:>]
>>
>> >>>> Sorry; my netiquette has gotten rusty it seems :(
>>
>> >>>>> Is this a 1x63 array? &#4294967295;3x21? &#4294967295;How many lamps do you intend to
>> >>>>> have lit concurrently? &#4294967295;how many "intensities" do you seek?
>>
>> >>>> The array configuration is one of the things that I'm uncertain on.
>> >>>> I would think that it'd be configured based on the end method of
>> >>>> driving said array. I'd need all on at the same time, with at least
>> >>>> a few intensity levels (5 or so would be plenty) for all LEDs.
>>
>> >>> Well, let's back of the envelope this. Figure a nominal 2 V drop per LED
>> >>> and you've got 5 V to work with so run two in series, 32 times. You want
>> >>> to PWM so figure a BJT switch with, say, 0.5 Vce at saturation, so a
>> >>> swag is 4.5 V per pair, and 32 pairs. (If you really want 63 and not 64,
>> >>> put some electrical tape over the 64th, or throw some regular diodes
>> >>> there in its place.)
>>
>> >>> For 500 mA max at the USB port, assuming a proper dialog with the port
>> >>> to get permission to draw more than 200 mA, you can push about 15 mA
>> >>> through each pair. Less, really, since the processor will want some and
>> >>> some will be lost in other ways. So, say you look for LEDs that are
>> >>> adequately bright at no more than 10 mA. A 47 ohm current limiting
>> >>> resistor for each pair might be a starting point.
>>
>> >> PIC32 (min 64 pins) has only 5 PWM outputs. &#4294967295;So, you need transistors
>> >> to buffer more than 18mA (max) per port. &#4294967295;PIC24 has 9 PWM, BTW.
>>
>> >If the OP is only interested in 5 light levels, you don't
>> >need a "real" PWM channel (do it "old school")
>
>Just one channel.  The issues were unclear before.
>
>>
>> [Crazy usenet. I can see posts from linnix timestamped before and after
>> that one but... not that one. %!#$! &#4294967295;::sigh:: ]
>>
>> Yes, I was imagining one PWM channel driving one transistor that then
>> tickled the bases of all of the 32 LED string transistors in parallel.
>> Rather a brute force approach, of course. Since they're operating as
>> saturation switches, minor differences in the transistor characteristics
>> should be mostly a wash.
>
>I was thinking just one tiny metal can (2N2222) that can handle 500mA
>collector current.  One of my first transistor and still as good as
>any.


Would work, but I'd be concerned with process variations between the
LEDs leading to the pair with the lowest forward voltage drop hogging
all of the current. Less of a problem (and likely a non-problem) if
they're all from the same batch, of course.

A current mirror is probably the best approach for balanced control.
Also opens the possibility of controlling the brightness by controlling
the current directly (and avoiding flicker or beat frequencies).

-- 
Rich Webb     Norfolk, VA

On 5/22/2011 5:39 PM, linnix wrote:
> On May 22, 5:17 pm, D Yuniskis<not.going.to...@seen.com>  wrote:

>> Wait -- you want to treat them all AS IF they were a single lamp?
>> I.e., two states:  all on (at some "dimness level") or all *off*?
>> *Not* 1 on, 2 on, 26 on, etc.?
>>
>> (which begs the question, why not use "some number on" to give you
>> your variable intensity?  unless you need illumination over a
>> wider area, etc.)
>
> Sound like LCD backlite.  It he said so originally, we wouldn't have
> to go through all those guessings.

<grin>  Welcome to USENET!

Hi Rich,

On 5/22/2011 6:14 PM, Rich Webb wrote:

>>>>> The array configuration is one of the things that I'm uncertain on.
>>>>> I would think that it'd be configured based on the end method of
>>>>> driving said array. I'd need all on at the same time, with at least
>>>>> a few intensity levels (5 or so would be plenty) for all LEDs.
>>>>
>>>> Well, let's back of the envelope this. Figure a nominal 2 V drop per LED
>>>> and you've got 5 V to work with so run two in series, 32 times. You want
>>>> to PWM so figure a BJT switch with, say, 0.5 Vce at saturation, so a
>>>> swag is 4.5 V per pair, and 32 pairs. (If you really want 63 and not 64,
>>>> put some electrical tape over the 64th, or throw some regular diodes
>>>> there in its place.)
>>>>
>>>> For 500 mA max at the USB port, assuming a proper dialog with the port
>>>> to get permission to draw more than 200 mA, you can push about 15 mA
>>>> through each pair. Less, really, since the processor will want some and
>>>> some will be lost in other ways. So, say you look for LEDs that are
>>>> adequately bright at no more than 10 mA. A 47 ohm current limiting
>>>> resistor for each pair might be a starting point.
>>>
>>> PIC32 (min 64 pins) has only 5 PWM outputs.  So, you need transistors
>>> to buffer more than 18mA (max) per port.  PIC24 has 9 PWM, BTW.
>>
>> If the OP is only interested in 5 light levels, you don't
>> need a "real" PWM channel (do it "old school")
>
> [Crazy usenet. I can see posts from linnix timestamped before and after
> that one but... not that one. %!#$!  ::sigh:: ]
>
> Yes, I was imagining one PWM channel driving one transistor that then
> tickled the bases of all of the 32 LED string transistors in parallel.
> Rather a brute force approach, of course. Since they're operating as
> saturation switches, minor differences in the transistor characteristics
> should be mostly a wash.

If the OP wants to treat them as *one* lamp, I would seriously
consider a small DC-DC converter to kick the 5V (assuming it wants
to be entirely powered off the USB) supply up to "whatever" @ 10mA.
Note that you could design the supply so that *it* does the intensity
control (current mode output that tracks something from the MCU).

I would be more concerned about having enough Icc from the USB
port to deliver what the lamps will consume.  (high efficiency
lamps?)

On May 22, 6:14=A0pm, Rich Webb <bbew...@mapson.nozirev.ten> wrote:
> On Sun, 22 May 2011 17:19:34 -0700, D Yuniskis
>
>
>
>
>
>
>
>
>
> <not.going.to...@seen.com> wrote:
> >On 5/22/2011 4:31 PM, linnix wrote:
> >> On May 22, 4:22 pm, Rich Webb<bbew...@mapson.nozirev.ten> =A0wrote:
> >>> On Sun, 22 May 2011 15:34:58 -0700 (PDT), Gene<profii...@gmail.com>
> >>> wrote:
>
> >>>> On Sunday, May 22, 2011 4:58:42 PM UTC-4, D Yuniskis wrote:
> >>>>> [grrr... how about learning to break your lines at something like
> >>>>> 70 characters?? =A0I've taken the liberty to *chop* off everything =
that
> >>>>> extends beyond the width of my window... no fun, eh? =A0:>]
>
> >>>> Sorry; my netiquette has gotten rusty it seems :(
>
> >>>>> Is this a 1x63 array? =A03x21? =A0How many lamps do you intend to
> >>>>> have lit concurrently? =A0how many "intensities" do you seek?
>
> >>>> The array configuration is one of the things that I'm uncertain on.
> >>>> I would think that it'd be configured based on the end method of
> >>>> driving said array. I'd need all on at the same time, with at least
> >>>> a few intensity levels (5 or so would be plenty) for all LEDs.
>
> >>> Well, let's back of the envelope this. Figure a nominal 2 V drop per =
LED
> >>> and you've got 5 V to work with so run two in series, 32 times. You w=
ant
> >>> to PWM so figure a BJT switch with, say, 0.5 Vce at saturation, so a
> >>> swag is 4.5 V per pair, and 32 pairs. (If you really want 63 and not =
64,
> >>> put some electrical tape over the 64th, or throw some regular diodes
> >>> there in its place.)
>
> >>> For 500 mA max at the USB port, assuming a proper dialog with the por=
t
> >>> to get permission to draw more than 200 mA, you can push about 15 mA
> >>> through each pair. Less, really, since the processor will want some a=
nd
> >>> some will be lost in other ways. So, say you look for LEDs that are
> >>> adequately bright at no more than 10 mA. A 47 ohm current limiting
> >>> resistor for each pair might be a starting point.
>
> >> PIC32 (min 64 pins) has only 5 PWM outputs. =A0So, you need transistor=
s
> >> to buffer more than 18mA (max) per port. =A0PIC24 has 9 PWM, BTW.
>
> >If the OP is only interested in 5 light levels, you don't
> >need a "real" PWM channel (do it "old school")

Just one channel.  The issues were unclear before.

>
> [Crazy usenet. I can see posts from linnix timestamped before and after
> that one but... not that one. %!#$! =A0::sigh:: ]
>
> Yes, I was imagining one PWM channel driving one transistor that then
> tickled the bases of all of the 32 LED string transistors in parallel.
> Rather a brute force approach, of course. Since they're operating as
> saturation switches, minor differences in the transistor characteristics
> should be mostly a wash.

I was thinking just one tiny metal can (2N2222) that can handle 500mA
collector current.  One of my first transistor and still as good as
any.

On Sun, 22 May 2011 17:19:34 -0700, D Yuniskis
<not.going.to.be@seen.com> wrote:

>On 5/22/2011 4:31 PM, linnix wrote:
>> On May 22, 4:22 pm, Rich Webb<bbew...@mapson.nozirev.ten>  wrote:
>>> On Sun, 22 May 2011 15:34:58 -0700 (PDT), Gene<profii...@gmail.com>
>>> wrote:
>>>
>>>> On Sunday, May 22, 2011 4:58:42 PM UTC-4, D Yuniskis wrote:
>>>>> [grrr... how about learning to break your lines at something like
>>>>> 70 characters??  I've taken the liberty to *chop* off everything that
>>>>> extends beyond the width of my window... no fun, eh?  :>]
>>>
>>>> Sorry; my netiquette has gotten rusty it seems :(
>>>
>>>>> Is this a 1x63 array?  3x21?  How many lamps do you intend to
>>>>> have lit concurrently?  how many "intensities" do you seek?
>>>
>>>> The array configuration is one of the things that I'm uncertain on.
>>>> I would think that it'd be configured based on the end method of
>>>> driving said array. I'd need all on at the same time, with at least
>>>> a few intensity levels (5 or so would be plenty) for all LEDs.
>>>
>>> Well, let's back of the envelope this. Figure a nominal 2 V drop per LED
>>> and you've got 5 V to work with so run two in series, 32 times. You want
>>> to PWM so figure a BJT switch with, say, 0.5 Vce at saturation, so a
>>> swag is 4.5 V per pair, and 32 pairs. (If you really want 63 and not 64,
>>> put some electrical tape over the 64th, or throw some regular diodes
>>> there in its place.)
>>>
>>> For 500 mA max at the USB port, assuming a proper dialog with the port
>>> to get permission to draw more than 200 mA, you can push about 15 mA
>>> through each pair. Less, really, since the processor will want some and
>>> some will be lost in other ways. So, say you look for LEDs that are
>>> adequately bright at no more than 10 mA. A 47 ohm current limiting
>>> resistor for each pair might be a starting point.
>>
>> PIC32 (min 64 pins) has only 5 PWM outputs.  So, you need transistors
>> to buffer more than 18mA (max) per port.  PIC24 has 9 PWM, BTW.
>
>If the OP is only interested in 5 light levels, you don't
>need a "real" PWM channel (do it "old school")

[Crazy usenet. I can see posts from linnix timestamped before and after
that one but... not that one. %!#$!  ::sigh:: ]

Yes, I was imagining one PWM channel driving one transistor that then
tickled the bases of all of the 32 LED string transistors in parallel.
Rather a brute force approach, of course. Since they're operating as
saturation switches, minor differences in the transistor characteristics
should be mostly a wash.

-- 
Rich Webb     Norfolk, VA

On 5/22/2011 6:19 PM, D Yuniskis wrote:
> On 5/22/2011 4:31 PM, linnix wrote:
>> On May 22, 4:22 pm, Rich Webb<bbew...@mapson.nozirev.ten> wrote:
>>> On Sun, 22 May 2011 15:34:58 -0700 (PDT), Gene<profii...@gmail.com>
>>> wrote:
>>>
>>>> On Sunday, May 22, 2011 4:58:42 PM UTC-4, D Yuniskis wrote:
>>>>> [grrr... how about learning to break your lines at something like
>>>>> 70 characters?? I've taken the liberty to *chop* off everything that
>>>>> extends beyond the width of my window... no fun, eh? :>]
>>>
>>>> Sorry; my netiquette has gotten rusty it seems :(
>>>
>>>>> Is this a 1x63 array? 3x21? How many lamps do you intend to
>>>>> have lit concurrently? how many "intensities" do you seek?
>>>
>>>> The array configuration is one of the things that I'm uncertain on.
>>>> I would think that it'd be configured based on the end method of
>>>> driving said array. I'd need all on at the same time, with at least
>>>> a few intensity levels (5 or so would be plenty) for all LEDs.
>>>
>>> Well, let's back of the envelope this. Figure a nominal 2 V drop per LED
>>> and you've got 5 V to work with so run two in series, 32 times. You want
>>> to PWM so figure a BJT switch with, say, 0.5 Vce at saturation, so a
>>> swag is 4.5 V per pair, and 32 pairs. (If you really want 63 and not 64,
>>> put some electrical tape over the 64th, or throw some regular diodes
>>> there in its place.)
>>>
>>> For 500 mA max at the USB port, assuming a proper dialog with the port
>>> to get permission to draw more than 200 mA, you can push about 15 mA
>>> through each pair. Less, really, since the processor will want some and
>>> some will be lost in other ways. So, say you look for LEDs that are
>>> adequately bright at no more than 10 mA. A 47 ohm current limiting
>>> resistor for each pair might be a starting point.
>>
>> PIC32 (min 64 pins) has only 5 PWM outputs. So, you need transistors
>> to buffer more than 18mA (max) per port. PIC24 has 9 PWM, BTW.
>
> If the OP is only interested in 5 light levels, you don't
> need a "real" PWM channel (do it "old school")

I too was wondering why so many PWM channels were required for a 
non-linear output.
Even a multiplexed array of diodes can get every LED a different level 
of intensity.
Just limit the number of levels, as D Yuniskis stated.

hamilton

On May 22, 5:10=A0pm, Gene <profii...@gmail.com> wrote:
> On Sunday, May 22, 2011 7:20:02 PM UTC-4, linnix wrote:
> > BTW, why PIC32? =A0There must be other reasons than just driving LEDs.
>
> It will also have a key matrix and a 16 bit lcd attached. There won't
> be a whole lot of pins left, ergo my thought about using a driver
> via i2c/spi or similar, perhaps.

Next time, try something like this:

Question:

What is the best way to drive 63 backlite LEDs for the LCD panel of a
PIC32,  with upto 5V 500mA USB bus power.

Answer:

PWM the base of a 2N2222, with 32 pairs of LEDs tied between collector
and a 10 ohms resistor from 5V VBUS.

On May 22, 5:17=A0pm, D Yuniskis <not.going.to...@seen.com> wrote:
> Hi Gene,
>
> On 5/22/2011 3:34 PM, Gene wrote:
>
> >> Is this a 1x63 array? =A03x21? =A0How many lamps do you intend to
> >> have lit concurrently? =A0how many "intensities" do you seek?
>
> > The array configuration is one of the things that I'm uncertain on.
> > I would think that it'd be configured based on the end method of
> > driving said array. I'd need all on at the same time, with at least
> > a few intensity levels (5 or so would be plenty) for all LEDs.
>
> Wait -- you want to treat them all AS IF they were a single lamp?
> I.e., two states: =A0all on (at some "dimness level") or all *off*?
> *Not* 1 on, 2 on, 26 on, etc.?
>
> (which begs the question, why not use "some number on" to give you
> your variable intensity? =A0unless you need illumination over a
> wider area, etc.)

Sound like LCD backlite.  It he said so originally, we wouldn't have
to go through all those guessings.