If there’s one thing we like, it’s blinky stuff, and you’re not going to get anything cooler than a display made of tiny SMD LEDs. That’s the idea behind this wristwatch and Hackaday Prize entry. It’s a tiny board, loaded up with an ATmega, a few buttons, and a bunch of LEDs in a big charlieplexed array.
The big feature of this display is the array of LEDs. This is a 16×5 array of 0603 LEDs packed together as tightly as possible. That’s a tiny, high-resolution LED display, but even with the ATmega88 microcontroller powering this board, all the LEDs are individually addressable, and a proper font for displaying the time, or anything else, is already mapped out.
LED matrices are pretty common around these parts, but building a custom display out of SMD LEDs is another level entirely. The best one we’ve seen was this unofficial badge from two DEF CONs ago. That was done the cheater’s way with a bunch of serially addressable LED drivers. This charlieplexed version goes above and beyond, and we’re eagerly awaiting the board files so this display can be replicated easily.
beautiful watches and absolute perfect designed PCB !!! I have something in round shape> https://www.tindie.com/products/bobricius/92-led-charlieplexed-ring-circle-watch-matrix/
Nice watch design you have there Peter. How are you driving the 4×5 LED matrix in the middle?
There is 2 independent charlieplex groups. …. Circle and matrix …. Matrix is only because I hate unused pins :)
I see in your tindie store that you are selling just the 4×5 charlieplexed matrix. The PCB layout can be made simpler on a single layer if you use more pins and the LEDs can be closer as vias would not be needed between them
This matrix is designed for attiny85 …. There is only 5 pins.
Well, I had a problem with LED charlieplexing in the past. This is, because the led breakdown voltage is near the light emitting voltage. Especially with cheap chinese red LED I had a visible burnout after about two months of continuous operation. The idea is great, but unless there are reverse protection diodes included in the design, I think it’s a bit problematic…
The LEDs I’m using have a rated reverse breakdown voltage of 5V and the battery being used is a CR2032 3V coin cell. So, I guess that would not be an issue
Vishay states for their 603 smd leds:
“Driving the LED in reverse direction is suitable for short term application”
http://www.vishay.com/docs/83172/tlmo1000.pdf Page 1 Bottom
And I found out while charlieplexing led stripes as ambient light, that there was a dead slow decreasing of luminuos intensity over two months of continuous operation. I just found out after completing a brand new stripe from the same lot of leds and it had a significant greater brightness and seeked for a reason. This reverse driving problem is known…
Yup. Ask anyone who bought a set of LED Christmas lights and expected them to outlast their old incandescent ones (Most LED Christmas lights let the LEDs act as AC rectifiers).
I wouldn’t be surprised if the Christmas light strings are overdriven in forward bias, too – every good designer knows that to get brighter output, you can just ignore the datasheet and just drive them harder. Who knows which effect is ultimately responsible for the string’s demise.
Not to mention that these LED are the cheapest grades from China with slack quality control to cut cost. Compounding the issue is that white ones use phosphor which degrades over time.
I did not know about the long term effect of reverse voltage. Thanks for pointing that out.
Even so, I don’t think it will be much of a problem because in my application, it is supposed to display the time/date when the button is pressed and then turn off after 5 seconds. There is also the screen refresh rate. So, effectively each LED would be reverse biased for a pretty short amount of time.
Wouldn’t the other LED connected n reverse clamps the voltage? i.e. reverse voltage is negative forward voltage of another LED.
https://cdn.hackaday.io/images/274141470842774905.png
There are pairs of LEDs that are connected in reverse for the same column: (row 1, row 2), (row 4, row 5). For row 3, it is between the neighbors (2n,2n+1).
The other member of the pair would act as reverse protection.
I think that the issue is that LEDs don’t handle being reverse biased well, and tend to burn out slowly when asked to do so.
This arrangement limits the reverse voltage instead of the full 3V or 5V that some think which is my point. Old red/green/amber etc would see around 2V while newer ones that are based on blue LED would see 3V or whatever the diode drop is.
The only way to find out for sure is to run a controlled experiment with one set pulsed and the other set pulsed with a reverse bias and see if one or both set suffers from urning out. (White LED has phosphors that can degrade over time.)
Excellent point tekkieneet. The LED in anti-parallel would limit the reverse voltage. So, need to do an experiment by applying the LED forward voltage across a reverse biased LED to see how long it survives and how its brightness degrades.
I have a LTspice testbench setup for the LED matrix. Will check the amount of time each LED experiences reverse voltage in a typical screen refresh. Also, will dig for more information on LED lifetime when reverse biased. Thanks for this information.