K
krw@att.bizzzzzzzzzzzz
Guest
On Fri, 10 Aug 2012 06:34:09 -0700 (PDT), fungus <tooby@artlum.com> wrote:
This week.
M
Michael A. Terrell
Guest
"krw@att.bizzzzzzzzzzzz" wrote:
No, there has been new listings there for over six months.
No, there has been new listings there for over six months.
F
fungus
Guest
On Friday, August 10, 2012 2:56:15 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:
There's no shortage of them on there...
Go to eBay and type "MAX7219".
There's no shortage of them on there...
K
krw@att.bizzzzzzzzzzzz
Guest
On Fri, 10 Aug 2012 07:09:44 -0700 (PDT), fungus <tooby@artlum.com> wrote:
a ballast resistor. A current source is simply a variable resistor.
You don't need 3.6V unless you want to put your eye out.
See above.
You CANNOT gain efficiency by boosting the voltage and then pissing it away in
a ballast resistor. A current source is simply a variable resistor.
Wrong.
No, because it's *NOT* boosted to your 5V.
No.
F
fungus
Guest
On Friday, August 10, 2012 2:54:55 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:
LED it'll work down to 4.2V before
the current drops.
That's not *quite* dead, ie. "marginal".
a) You have to stick an extra battery in.
b) Your BJT+MOSFET (or whatever) is acting
exactly like a ballast resistor. It will
use the same power.
To increase the efficiency you'd have to
use a PWM circuit which produces an *average*
current without consuming much power itself.
ie. You need 4.2 volts - 3xAA isn't enough
With a 0.6V overhead and a 3.6V
LED it'll work down to 4.2V before
the current drops.
That's not *quite* dead, ie. "marginal".
How?
a) You have to stick an extra battery in.
b) Your BJT+MOSFET (or whatever) is acting
exactly like a ballast resistor. It will
use the same power.
To increase the efficiency you'd have to
use a PWM circuit which produces an *average*
current without consuming much power itself.
F
fungus
Guest
On Friday, August 10, 2012 6:03:37 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:
the 3.6V LED at 3.6V. Got it.
In that case I'm going to run mine at 2.5V.
I can connect two of them in series and get
over 90% efficiency. Let's see if your current
source can beat that!
with an extra battery your circuit will
have a fairly similar efficiency to my
booster-board circuit.
If you're not adding a battery then you
obviously win unless I can find a 3.6V
booster.
Ahhhh, your argument is based on not running
the 3.6V LED at 3.6V. Got it.
In that case I'm going to run mine at 2.5V.
I can connect two of them in series and get
over 90% efficiency. Let's see if your current
source can beat that!
I'm not saying you can. I'm saying that
with an extra battery your circuit will
have a fairly similar efficiency to my
booster-board circuit.
If you're not adding a battery then you
obviously win unless I can find a 3.6V
booster.
No argument there.
K
krw@att.bizzzzzzzzzzzz
Guest
On Fri, 10 Aug 2012 09:49:38 -0700 (PDT), fungus <tooby@artlum.com> wrote:
it's *not* needed. It'll be quite visible at 1/4 (or 1/10) of its rated
current. At 1/4, we found that blues were quite visible in full sunlight.
Running them "hotter" is nuts.
A 3.6V LED is only 3.6V at its rated current. Unless you're lighting a room,
it's *not* needed. It'll be quite visible at 1/4 (or 1/10) of its rated
current. At 1/4, we found that blues were quite visible in full sunlight.
Running them "hotter" is nuts.
We now know you're an idiot.
Utter nonsense.
You're still paying 11% for the boost.
D
Daniel Pitts
Guest
On 8/10/12 3:13 AM, fungus wrote:
I got my TL5916 for 88 cents, and the 74HC238 for 25 cents. Where the
MAX7219 is over $10 from the same supplier ($13 if I wanted DIPS, which
I do for now).
So to power 192 LEDs (8x8xRGB) I could spend spend over $30 using
MAX7219, or I could spend just 3*.88+.25=$2.89. I think a 90% cheaper
solution is the better solution, especially for a hobby project.
I recall, the MAX7219 was much more expensive for what I needed.
I got my TL5916 for 88 cents, and the 74HC238 for 25 cents. Where the
MAX7219 is over $10 from the same supplier ($13 if I wanted DIPS, which
I do for now).
So to power 192 LEDs (8x8xRGB) I could spend spend over $30 using
MAX7219, or I could spend just 3*.88+.25=$2.89. I think a 90% cheaper
solution is the better solution, especially for a hobby project.
F
fungus
Guest
On Friday, August 10, 2012 9:06:13 PM UTC+2, k...@att.bizzzzzzzzzzzz wrote:
you see the LED?"
and arbitrary LED brightness that just happens
to support your claim.
If we're allowed to pick and choose then I'll
get an adjustable boost board, eg:
http://www.ebay.com/itm/130736597148
I can use that to one with 2xAA batteries.
It uses PWM switching to do its job, I bet
it's more efficient for lighting an LED than
3xAA plus a current source that's guaranteed
to eat 0.6V.
Says who? Not all LED applications are "can
you see the LED?"
Me? You're the one picking a particular voltage
and arbitrary LED brightness that just happens
to support your claim.
If we're allowed to pick and choose then I'll
get an adjustable boost board, eg:
http://www.ebay.com/itm/130736597148
I can use that to one with 2xAA batteries.
It uses PWM switching to do its job, I bet
it's more efficient for lighting an LED than
3xAA plus a current source that's guaranteed
to eat 0.6V.
K
krw@att.bizzzzzzzzzzzz
Guest
On Fri, 10 Aug 2012 13:19:53 -0700 (PDT), fungus <tooby@artlum.com> wrote:
of an ass of yourself.
really on Obama's campaign team?
with it.
You don't want to see the LED? Read the fucking thread before you make more
of an ass of yourself.
You really are trying to be right at all costs, even by being wrong. Are you
really on Obama's campaign team?
Of course you would. You've already demonstrated that you're clueless.
If you're going to all that trouble, just use a damned wall wart and be done
with it.
F
fungus
Guest
On Friday, August 10, 2012 11:34:58 PM UTC+2, Daniel Pitts wrote:
http://www.ebay.com/itm/280849405928
Or ... two of them for $1.75:
http://www.ebay.com/itm/300739928063
The only thing I didn't figure out yet is
scalability. The MAX7219 is perfect for a
4x4x4 cube. Two of them can do a 5x5x5 cube
but do I need eight of them for an 8x8x8...?
Can I add a multiplexer and do it with just
one chip? It might work, I dunno. The
multiplexer runs at 800Hz. Is it enough...?
The TL5916 *might* be more scalable because
you have total control over all the timing.
I'm just reading all the datasheets now,
thinking it might be a good time to get
started on my own LED cube.
On eBay they go for $0.99 with shipping included...
http://www.ebay.com/itm/280849405928
Or ... two of them for $1.75:
http://www.ebay.com/itm/300739928063
The only thing I didn't figure out yet is
scalability. The MAX7219 is perfect for a
4x4x4 cube. Two of them can do a 5x5x5 cube
but do I need eight of them for an 8x8x8...?
Can I add a multiplexer and do it with just
one chip? It might work, I dunno. The
multiplexer runs at 800Hz. Is it enough...?
The TL5916 *might* be more scalable because
you have total control over all the timing.
I'm just reading all the datasheets now,
thinking it might be a good time to get
started on my own LED cube.
F
fungus
Guest
On Saturday, August 11, 2012 1:05:57 AM UTC+2, fungus wrote:
two MAX7219s can light up 128 LEDs, they have
be connected in 8x8 matrices. Physically wiring
up a 5x5x5 cube using two chips isn't going to
work out. You could theoretically do it with
three chips but it's messy, you can't build it
in five neat slices then stack them up. Saving
a couple of bucks to use three chips instead
of five doesn't seem worth it for a hobby project.
4x4x4 can be done neatly with one chip but after
that you're really looking at one chip per layer.
PS: Odd-numbered cubes seem to be much more
interesting/pretty then even-numbered cubes.
5x5x5 and 7x7x7 are good sizes. 7x7x7 is
better than 8x8x8.
I found a big flaw in my thinking: Although
two MAX7219s can light up 128 LEDs, they have
be connected in 8x8 matrices. Physically wiring
up a 5x5x5 cube using two chips isn't going to
work out. You could theoretically do it with
three chips but it's messy, you can't build it
in five neat slices then stack them up. Saving
a couple of bucks to use three chips instead
of five doesn't seem worth it for a hobby project.
4x4x4 can be done neatly with one chip but after
that you're really looking at one chip per layer.
PS: Odd-numbered cubes seem to be much more
interesting/pretty then even-numbered cubes.
5x5x5 and 7x7x7 are good sizes. 7x7x7 is
better than 8x8x8.
F
fungus
Guest
On Saturday, August 11, 2012 1:05:57 AM UTC+2, fungus wrote:
http://www.ebay.com/itm/300739928460
Now I just need a load of LEDs and my
cube will be under way...
I think I'll go for 5x5x5 - 7x7x7 is
three times as many LEDs, that's a lot
more work/expense than 5x5x5.
I gave in to temptation, ten MAX7219s for $5.10thinking it might be a good time to get
started on my own LED cube.
http://www.ebay.com/itm/300739928460
Now I just need a load of LEDs and my
cube will be under way...
I think I'll go for 5x5x5 - 7x7x7 is
three times as many LEDs, that's a lot
more work/expense than 5x5x5.
J
Jon Kirwan
Guest
On Sun, 12 Aug 2012 03:52:01 -0700 (PDT), fungus
<tooby@artlum.com> wrote:
ICs for driving all that. Separate supplies for each color,
turn-pot resistors for adjusting white balance by setting the
100% current source levels, and pwm capability within that
range from 0/256 to 255/256 of the 100% current level for
each color of each RGB LED. Used mostly to make outdoor TV
sets, driven by NTSC or some other TV broadcast code.
These bricks have huge heat sinks behind each one to
dissipate the power. Be aware of the potentials for that
aspect when you start talking about hundreds of LEDs.
Jon
<tooby@artlum.com> wrote:
I've got some very nice, 16x16 RGB arrays. Each already has 6On Saturday, August 11, 2012 1:05:57 AM UTC+2, fungus wrote:
thinking it might be a good time to get
started on my own LED cube.
I gave in to temptation, ten MAX7219s for $5.10
http://www.ebay.com/itm/300739928460
Now I just need a load of LEDs and my
cube will be under way...
I think I'll go for 5x5x5 - 7x7x7 is
three times as many LEDs, that's a lot
more work/expense than 5x5x5.
ICs for driving all that. Separate supplies for each color,
turn-pot resistors for adjusting white balance by setting the
100% current source levels, and pwm capability within that
range from 0/256 to 255/256 of the 100% current level for
each color of each RGB LED. Used mostly to make outdoor TV
sets, driven by NTSC or some other TV broadcast code.
These bricks have huge heat sinks behind each one to
dissipate the power. Be aware of the potentials for that
aspect when you start talking about hundreds of LEDs.
Jon
F
fungus
Guest
On Saturday, August 11, 2012 12:47:45 AM UTC+2, k...@att.bizzzzzzzzzzzz wrote:
You might just want the light from it.
I'm sure you can think of an example if
you really try.
Methinks you're trying to steer attention
away from your untenable position.
You might not want to see the LED directly, no.
You might just want the light from it.
I'm sure you can think of an example if
you really try.
How is that any 'trouble'?
Methinks you're trying to steer attention
away from your untenable position.
F
fungus
Guest
On Sunday, August 12, 2012 11:42:40 PM UTC+2, Jon Kirwan wrote:
the LEDs have more air around them.
How are they multiplexed? What current do the LEDs
run at? I imagine they drive them quite hard for
daylight viewing.
They're packed together quite tightly. In a cube
the LEDs have more air around them.
How are they multiplexed? What current do the LEDs
run at? I imagine they drive them quite hard for
daylight viewing.
K
krw@att.bizzzzzzzzzzzz
Guest
On Sun, 12 Aug 2012 17:30:03 -0700 (PDT), fungus <tooby@artlum.com> wrote:
READ THE THREAD, MORON.
I bet you could understand a simple thread if you really worked at it.
Idiot.
Look in the fucking mirror, moron.
J
Jon Kirwan
Guest
On Sun, 12 Aug 2012 18:10:24 -0700 (PDT), fungus
<tooby@artlum.com> wrote:
halves. Each half has 3 driver ICs, one for each color. Each
with different source supplies. (Blue needs higher voltage
than red, for example, and at these dissipations you do NOT
want the same rail for all three colors.)
dissipate about 80W per.
Jon
<tooby@artlum.com> wrote:
Well, their separation is 5mm and 3mm, depending on module.
There are six driver ICs. The 16x16 is divided into two 8x16
halves. Each half has 3 driver ICs, one for each color. Each
with different source supplies. (Blue needs higher voltage
than red, for example, and at these dissipations you do NOT
want the same rail for all three colors.)
I'd have to check my notes. But I remember that they roughly
dissipate about 80W per.
Jon
J
Jon Kirwan
Guest
On Mon, 13 Aug 2012 03:34:40 -0700 (PDT), fungus
<tooby@artlum.com> wrote:
chips. Each LED accepts 20mA, but at 3V, 3V, and 2V. So you
are talking about 160mW per, if on full. That doesn't count
any losses anywhere else supplying that. With 64, that's
about 10W. My panels are 4X as much. So that would be 40W
already. And you still don't have the drivers and their
losses. And besides, these LED panels I have are very high
efficiency and designed for outdoors so quite bright.
Regardless, you can see that the 80W figure is probably in
the ballpark. So my memory may have been correct. I think the
peak current was around 25mA for each LED.
Each IC was a 76 pad, 16x8 pixel controller with on-chip
static ram, EEPROM, Address Decoders, Multiplex circuitry and
constant current driver. It used 0.5um double metal, double
poly, CMOS technology.
Some features of these ICs were:
8 rows x 16 columns driver matrix.
Inter-digit blanking of 1/32 times the row-on period.
Programmable column driver constant current sources.
Up to 31 Levels of PWM global brightness control.
7 Bit global peak current digital brightness trim
in non-volatile memory.
7 Bit peak current global color correction from a
second color in non-volatile memory.
1024 values of programmable pre-scale for the
multiplex clock.
Internally buffered signal feed-through for serial
cascading of devices
24 Bit pulse counter (run time) in NV memory with
selectable 50 or 60Hz signal input.
70 MHz maximum serial shift clock and multiplex
clock frequency
30ms Deadman timer feature to prevent extended
direct current drive
130C Over-temp protection.
Half-multiplex (1/8 to 1/4) feature to decrease
peak current.
Programmable stagger of column drivers for reduced EMI
Pretty fancy, all in all.
All this was built into a module with a metal, drilled out
face for each RGB LED and a heavy heat sink on the backside.
You supplied power and serial communications.
It would take some serious design, microcontrollers, and four
of those RGB 8x8 on Ebay (they seem to got for about $7-8
each) to put the equivalent together. They sold, in qty, for
about $70 each. And that was some years ago. They are OSRAM
devices.
And yes, a 200x300 matrix (VHS supported 210 lines) would be
about 250 devices or something like that. I think the
displays rarely had all the pixels full-on all the time, so I
think 10kW was the area folks considered for design. It was
still a LOT of power. So heat (and the communication rate and
distribution to keep 30 frames per second going) were the
main issues, I think.
Jon
<tooby@artlum.com> wrote:
Well, if you look at Ebay you will see 8x8 RGB without driver
chips. Each LED accepts 20mA, but at 3V, 3V, and 2V. So you
are talking about 160mW per, if on full. That doesn't count
any losses anywhere else supplying that. With 64, that's
about 10W. My panels are 4X as much. So that would be 40W
already. And you still don't have the drivers and their
losses. And besides, these LED panels I have are very high
efficiency and designed for outdoors so quite bright.
Regardless, you can see that the 80W figure is probably in
the ballpark. So my memory may have been correct. I think the
peak current was around 25mA for each LED.
Each IC was a 76 pad, 16x8 pixel controller with on-chip
static ram, EEPROM, Address Decoders, Multiplex circuitry and
constant current driver. It used 0.5um double metal, double
poly, CMOS technology.
Some features of these ICs were:
8 rows x 16 columns driver matrix.
Inter-digit blanking of 1/32 times the row-on period.
Programmable column driver constant current sources.
Up to 31 Levels of PWM global brightness control.
7 Bit global peak current digital brightness trim
in non-volatile memory.
7 Bit peak current global color correction from a
second color in non-volatile memory.
1024 values of programmable pre-scale for the
multiplex clock.
Internally buffered signal feed-through for serial
cascading of devices
24 Bit pulse counter (run time) in NV memory with
selectable 50 or 60Hz signal input.
70 MHz maximum serial shift clock and multiplex
clock frequency
30ms Deadman timer feature to prevent extended
direct current drive
130C Over-temp protection.
Half-multiplex (1/8 to 1/4) feature to decrease
peak current.
Programmable stagger of column drivers for reduced EMI
Pretty fancy, all in all.
All this was built into a module with a metal, drilled out
face for each RGB LED and a heavy heat sink on the backside.
You supplied power and serial communications.
It would take some serious design, microcontrollers, and four
of those RGB 8x8 on Ebay (they seem to got for about $7-8
each) to put the equivalent together. They sold, in qty, for
about $70 each. And that was some years ago. They are OSRAM
devices.
And yes, a 200x300 matrix (VHS supported 210 lines) would be
about 250 devices or something like that. I think the
displays rarely had all the pixels full-on all the time, so I
think 10kW was the area folks considered for design. It was
still a LOT of power. So heat (and the communication rate and
distribution to keep 30 frames per second going) were the
main issues, I think.
Jon
J
Jon Kirwan
Guest
Here's a picture of the devices:
http://www.infinitefactors.org/misc/images/p1000046_640x480.jpg
3mm diameter pixel on a 4mm pitch, with a wide viewing angle.
Jon
http://www.infinitefactors.org/misc/images/p1000046_640x480.jpg
3mm diameter pixel on a 4mm pitch, with a wide viewing angle.
Jon