weird red LED case replacment

[lamoule74]

Hey there, I'm new on the eevblog forum,

I was intended to replace a 7 segments display on a scale, but when I disassembled it I've been surprised.

It seems to be composed by smalls "taped" LEDs... Never seen that.
I've no idea how to replace that.

Does any know how to replace it ?

Cheers.

[Twoflower]

That looks like Chip On Board (COB). So they put the individual, unpacked LED dies on the PCB and bonded them. Not easy to replace them 1:1. Unless you have access to a bonding machine.

If you're lucky you can place small SMDs in place of the die. But that would require more space between the PCB and the diffuser. And they might light up the segments not that evenly. If not replace that whole board with a self made PCB with stock 7-Segment modules. Again if the space allows.

[inse]

The layout is really odd.
Are there contacts in the diffuser as I see pads for jumpers crossing other traces.
And the traces that are interrupted by drill holes, strange..
I cannot see bond wires in your macro photos, are they broken off or simply not visible?
Did you check the LEDs with a multimeter?
I recently modified a LED backlight for a small LCD which had PCB strips with bonded LED chips.
I scratched off the remains and soldered really tiny LEDs (0603) on the pads just for boredom and to put the LEDs to use.
They were a find in the waste bin.

[james_s]

That's very common, or used to be anyway. It was the standard way of making displays for digital alarm clocks in the 80s-90s, I don't know if they still make them that way. As someone already said it's just bare LED dies bonded to the PCB.

[Decapitator]

It looks like an old multiplexed 4 digit display. How many segments are out? Are one or more digits totally dead? Is the same segment out on all 4 digits? If so it could be a driver chip or connection issue.

It may be possible to replace the entire LED board with something of your own creation with common 7 segment displays if you have enough room. You would need to determine whether to use common anode or common cathode.

[lamoule74]

Hello world,

Thanks for all these replies.

There were only one segment off,
I tried to turn the LEDs ON with my laboratory power supply set at 2V + 2 probes...
It worked for only 2 LEDs, then nothing...
I was a bit circumspect and after a matter of minutes so I decided to reassemble the scale.

Then none of the segments could worked. Incredible, I don't see what I could did wrong...

At one moment, when I used microscope to inspect the PCB, it seemed to me to perceive a thin wire...
I probably scratched it and turn it down, but I didn't do that with all the LEDs !

Anyway, this repair lasted too long.
The scale remains to the bin, a lots of repairs waiting for me.
Rest in peace 80's scale.

And thank you for your help gentlemens.

[Haenk]

Just a quick note - this is not a thing of the 80s, but rather 70s. I would assume the "Bowmar Optostic" pocket calculator display was the earliest use of COB LEDs and one large plastic lens, shortly followed by National Semiconductor (in the Novus 650).
However those LEDs were tiny and in no way usable for a large display.

[Gyro]

...
There were only one segment off,
I tried to turn the LEDs ON with my laboratory power supply set at 2V + 2 probes...
It worked for only 2 LEDs, then nothing...
...

Just a note that this isn't a good way of testing LEDs, they (like most things) have a tolerance range. If you drive them with a constant 2V from a reasonably beefy source (a lab PSU) then the ones with a Vf of, say, 1.8V might get excessive current, while ones with a Vf of 2.1V might not show any visible light.

You are far better to set the PSU to maybe 3V, and insert a 100R series resistor on one of your probes. This will limit the current through a typical Red LED to around 10mA, or 30mA if you a short the probes. Of course you can go higher, for instance 5V, with a correspondingly higher 330R series resistor, to ensure that other LED colours light too.

The practical upper limit is to keep the voltage below the typical reverse breakdown voltage of an LED, to avoid damage if you get the polarity wrong.

[lamoule74]

Just a note that this isn't a good way of testing LEDs, they (like most things) have a tolerance range. If you drive them with a constant 2V from a reasonably beefy source (a lab PSU) then the ones with a Vf of, say, 1.8V might get excessive current, while ones with a Vf of 2.1V might not show any visible light.

Right,
I'm also using the LED-test on the multimeter,
dunno why I change my way that tie...

Sorry for the grand chelem btw