Dead LED

[Bigglez]

I dont have any specific info on that chip

Have you followed the link to a datasheet that I posted earlier?

Another possibility is that the reverse voltage when the LED
is 'off' is too high. This is something that could easily be
checked with a scope.

LED reverse voltage spec is 5V, power to the PCB is 5.159V

I checked the bad diode's drive waveforms, comparing them
to an adjacent LED. Then I replaced the bad LED and checked again.

I also compared the bad LED with a good one off the board,
using a DMM on the diode range:
Bad LED was 0.201 A to K and K to A
Ref LED was 1.801 A to K and 'OL' K to A
I conclude the bad LED was shorted.

Here's the scope PIX:

[MrAl]

Hi again,


OH ok, i followed the link and took a quick look at the ic now.
I see it is a segment scanner with SPI interface.

What i still dont know however is what the startup of this chip
looks like. I assume you are using a micro controller just
ahead of the scanner chip? Depending on how long the micro
takes to start up i would wonder what the segment scanner
chip is doing...is it sitting there idle or is it banging one of the
segments with heavy current?
In the old days you could actually see this happening with some
devices. You would turn them one, see one segment light up
real bright, then it would switch to 'normal' operation. If you
could devise a test for that possible scenario that might help.

Also, we have to keep in mind that we may be jumping to
conclusions here. It could have simply been a faulty LED
for any of a number of reasons. Perhaps we should wait
and see if another blows out relatively soon. Perhaps
all this 'worry' is for naught.
Perhaps take the time it took to burn out that one LED and
double it, then wait that time to see if another burns out.
If it does, i'd suspect something going wrong. If not, that's
pretty much that

[Bigglez]

What i still dont know however is what the startup of this chip
looks like.

Shutdown Mode
When the MAX7219 is in shutdown mode, the scan oscillator
is halted, all segment current sources are pulled to
ground, and all digit drivers are pulled to V+, thereby
blanking the display.

And

Typically, it takes less than 250μs for the MAX7219/
MAX7221 to leave shutdown mode.

I assume you are using a micro controller just
ahead of the scanner chip?

Yes, for my version. The PCB was designed for the dual role
of using MAXIM's PC software support over a parallel port.
I have used this PCB in both modes, but currently they
operate in standalone, without PC support.

Also, we have to keep in mind that we may be jumping to
conclusions here. It could have simply been a faulty LED
for any of a number of reasons. Perhaps we should wait
and see if another blows out relatively soon. Perhaps
all this 'worry' is for naught.

Agreed.

[MrAl]

Hi again,

Sounds good so far. Im glad you pointed that stuff out
without making me read the whole data sheet and study
the whole problem completely.

What would really satisfy my curiosity would be if you
could obtain a current measurement of the ic chip as
the power was turned on. Since the chip is in complete
control of the current i would think a measurement of
the supply input pin would work out well. This would
cover a lot of bases, including unlucky program timing
during startup. This current measurement would have
to exclude the bypass capacitor of course, and it would
have to be done with a scope so we could see the spike,
if any, during the first few instants after power 'on'.
I would bet the program code starts up in 50us or less.
That is, the time between power 'on' and the time the
first code is sent to the driver chip and enabled.

If this were my project and i was only building one or
two pieces however i would probably wait to see if
another LED burnt out before doing anything else,
even measurements, unless they were easy enough
to perform.

[jollyrgr]

I see LED lifespans listed in ten of thousands of hours and into the one hundred thousand hour life. What is interesting is the observation of some LED clock radios. The colon indicator on these are lit 24/7 when plugged in. One radio I have has been in use for 20 or so years. With the exception of power failures it is always plugged in and running. Rough calculation says this is at 175,000 plus hours of operation. Not as good as that lightbulb over 100 years old that is still burning but still interesting.

[MrAl]

Hi Jolly,


Any lightbulb that has been running for 100 years has to have been
manufactured at least 100 year ago. That's the only way it can
run for 100 years
Anything made today can run for 100 minutes, if lucky ha ha.

Ok, seriously, 100 days if you are lucky.

[Bigglez]

Not as good as that lightbulb over 100 years old that is still burning but still interesting.

This one? Funny story, I have a Shelby Lamp Works
antique bulb in my collection, at least I think it is. As I knew
little about it I did some Google research and discovered the 100 year
old bulb is also a Shelby (I think), and not too far away (Livermore
Ca.)

A friend and I went to see it. But it wasn't where we were told!
Turns out the centennial bulb outlasted the building and
had to be moved down the street years earlier...

[Bigglez]

I see LED lifespans listed in ten of thousands of hours and into the one hundred thousand hour life.

I routinely return CF lamps to Home Despot for refunds
or exchanges. They don't last seven years! I'm in the
habit of putting the dates on all new light bulbs (not just
CFs) to see how long they play, and keeping the receipts.

Last week my oldest one died. Put it in roughly a month
after we bought the home, and I by-passed the switch
so that it is on 24/7 as an outside night light and daytime
yellow CF bug light.

Let's see, June 2001 to September 2008... Wow, one actually
made it seven years! I can't take it back with a straight face.

[MrAl]

Hi again,


He he, the fine print on many of those bulbs read that the lifetime
is based on use of 8 hours per day...ha ha. I checked several
from Home Depot myself.

[CeaSaR]

Bigglez wrote:

This one?

That's the one posted here. Pretty amazing, huh?

CeaSaR