My problem is that some leds keep burning. Most of the time, the burned leds are the top row whitch are always ON "the time is" but this can be random.
Is it that the leds are cheap from ebay? I buy them in tape of 1000pcs.
Is it that I use wrong resistor for the MAX7219? I made a lot of different tests from 22.6K and up to 30K.
Is it that the MAX7219 chips are cheap from ebay?
Is it that the leds are cheap from ebay? I buy them in tape of 1000pcs.
Maybe. What are the specifications for the LEDs? Forward Voltage and max current are important. The first you can test for with a multimeter, the second depends on the batch of LEDs if you are not buying branded components.
Is it that I use wrong resistor for the MAX7219? I made a lot of different tests from 22.6K and up to 30K.
That is still around 20mA per LED. Try higher values for Rset.
Is it that the MAX7219 chips are cheap from ebay?
I've never had any problems with the 'alternative supply'.
How have you distributed the LEDs between the different MAX7219 ICs? Have you shut down the 'digits' that you don't use? This increases the refresh rate on the remaining LEDs and may cause problems if you are driving them too hard.
Is it that I use wrong resistor for the MAX7219? I made a lot of different tests from 22.6K and up to 30K.
Is it that the MAX7219 chips are cheap from ebay?
The resistor values sound quite conservative. I used 10K for all the 4x4x4 led cubes I built. The green one is still ok. With the blue one, the leds burned. I replaced them with red leds and they have been ok. The blue leds were cheap from eBay, the red and green were better quality from a more reliable supplier, I believe. It is possible the non-genuine max7219 chips are the problem but I guess probably not. I used similar cloned chips for my cubes. Mike is probably correct, I would agree.
marco_c:
Maybe. What are the specifications for the LEDs? Forward Voltage and max current are important. The first you can test for with a multimeter, the second depends on the batch of LEDs if you are not buying branded components.
That is still around 20mA per LED. Try higher values for Rset.
I've never had any problems with the 'alternative supply'.
From the specs of the ebay description usually says
If=20mA
Forward Voltage(V) Min 3.2
Forward Voltage(V) Typ 3.4
PaulRB:
The resistor values sound quite conservative. I used 10K for all the 4x4x4 led cubes I built.
You use even smaller resistor than I use. To play in the safe zone is better to increase the Rset or not? What is a safe value? I don't mind reducing the brightness a bit. Should I go for 50K for example?
I already use the "setIntensity" to set the brightness to 10 from the 0 to 15 range.
marco_c:
How have you distributed the LEDs between the different MAX7219 ICs? Have you shut down the 'digits' that you don't use? This increases the refresh rate on the remaining LEDs and may cause problems if you are driving them too hard.
I use it as 7x7 matrix.
With "setScanLimit" I exclude the last row since I don't use it. Is it wrong?
I am also trying to find bright white 3528 SMD leds from known suppliers with no luck since the cost is really high or they don't have. From digikey or farnell.
Any suggestions?
increase Rset. To how much? Say 50K?
from coding side is there something to improve?
since the clock needs 198 leds what alternative suppliers could I use?
The USB power pack provides the 5V directly to the "control board". Is that safe? Should I use any Caps (like I use for the MAX7219 chips) or any regulator?
22K is about right according to the data sheet (20mA, 3.5Vff)
Nothing in your code can burn the leds. If you were directly multiplexing the leds with the Arduino then perhaps. But you are using the max7219 chips, which are designed for the purpose of driving leds.
For 200 leds you should get a reasonable price from one of the big component suppliers.
More caps can't hurt. You can't use a regulator because your psu is 5V and your circuit needs 5V. A regulator would not improve things even with a higher voltage psu. Not sure why you are using a usb power pack. Won't last long I suspect.
I can't see anything else you are doing wrong, so I still think you must have had some poor quality leds.
10c per led does not sound too bad to me, for a hobby project. Just think of the value of your time that has gone into the project, as if you were paying yourself minimum wage.
mirmigki:
My problem is that some leds keep burning. Most of the time, the burned leds are the top row whitch are always ON "the time is" but this can be random.
Sub-standard components may well be a factor. But have you considered heat dissipation? The specs for current are at a nominal temperature, typically 25 deg. There will be a derating curve, for some similar LEDs I looked at, it might be 50% at 65 deg (i.e. instead of 20mA, the limit might be 10mA).
If you are in Greece, maybe you have a high ambient temperature?
It looks like your LEDs are in a fairly well insulated box, so the temperature, especially at the top, may well be a factor. If you can get a thermocouple in there to measure the temperature, it would be useful to know.
Certainly, keeping LEDs cool will help them last longer. I would put copper areas on PCB front and back, with several vias to join the sides. You may have them already, but I would put some ventilation slots in the box to get some airflow.
Hi George , did you ever resolve the burning LED issue ?
I’ve made a word clock also very similar to your method using smd components
I have tried LEDs from 3 different suppliers and even soak tested the same LEDs by slightly overdriving them to see if they burn out without any problems . However when used in the clock I get random ones burn out .
I’m putting it down to the multiplexing from the max7219 chip and the reverse bias to the led when in the off state ( the led receives 5 v in the reverse direction )
I’ve tested the LEDs upto 20vdc in reverse with only uA been passed , not enough to break down the led
Most of the data sheets on smd LEDs don’t reference much to the breakdown voltage in reverse bias and the ones that do only lists upto 5v . I’m assuming not many applications perform reverse biasing
I’m a little stumped . I have tried blocking diodes on the cathode output of the max7219 but this causing issues with random LEDs turn on.
Hopefully you or someone might have some answers other than use different LEDs
ok , thats what i assumed you meant , been at this for a while now and its kinda doing my head in ...
very temped to go down the path of trying out with the older Max7221 with dosnt reverse bias just in case it was that.. possibly might be due to the disruption of the oscillator from the PWM power supply I have in place ( simple buck dc to dc regulator ) may have to try better caps?
That might be the problem. These things are notorious for stopping working at certain currents. They are very sensitive to layout. I used to manage specialist PSU engineers and switch mode supplies were never completed in less than three CAD spins ( attempts at the layout ). The boost ones are especially prone to this.
It might be better to do a Pi filter ( the last circuit on my decoupling page ) rather than just use decoupling capacitors because this is high frequency noise. http://www.thebox.myzen.co.uk/Tutorial/De-coupling.html
Having said that the MAX7219 chip is expensive and so most of us resort to getting it from China, where the quality is suspect. Maybe they are clones or maybe from the reject bin.
I actually have a few Pi filters boards from a job that they didnt get used on for exactly that reason ( they incorporated 15A Buck Boosts ) job was redesigned , will jerry rig it into the assembly and test out
Much like the layout of your last image
I thought about the clone Max7219 been the issue , so i sampled a handful from Maxim to make sure .
the first build was all genuine parts as it turned out , the second was from cheaper clones ( ICs )
Both builds do the same thing with the LED's dropping out
Ill give you suggestion a go and see how it goes
