LED Cube Verticle Drawing Failure

[benedpd]

Hi everyone,

I successfully completed my 8x8x8 LED cube (by CHR) (https://www.instructables.com/Led-Cube-8x8x8/) some months ago. The vague idea is that an ATMEGA32 sends data to shift registers which then store it and choose which LEDs to give the most percentage on time in the PWM signal sent to the column (apologies for absolutely botching this).

Recently decided I wanted to make it more portable by simply adding a 3xAA battery pack to power it. However, in the process of completing that (taking the boards out of the casing and adding in the battery pack to the power lines), my cube has now started running into issues. The issues may have occurred before this changeover but not that I was aware of. I have attached schematics of the boards below.

In particular, the cube appears to struggle to draw patterns upwards or downwards when lighting up more than a few columns. Roughly 90% of the animations from CHR (original instructable poster) and SuperTech-IT (a second 8x8x8 LED cube instructable poster https://www.instructables.com/CHRs-8X8X8-LED-Cube-Revisited-with-improvements/) work flawlessly.

The first video shows a diagnostic test. It shows that each column can be lit up without issue. The diagnostic is then meant to light up each and every LED one by one from the bottom. However, mine instead appears to light up each layer (via each LED) one by one, but the last verticle plane always fails to do so. In the end the whole cube is meant to be lit.

In the second video, it shows more clearly the issue my cube has with lighting up each of the horizontal layers one by one. This is shown when it looks like a single column lights up dimly. It also shows how other (potentially less power consuming) animations work flawlessly.

These tests make me vaguely think it could be something to do with the transistor layer select array however I am not sure (the instructable used PN2222 transistors as it was what they had laying around, I followed this as I too had them lying around). Any advice would be beyond appreciated.

Thanks

 

[Diver300]

You are lighting up to 64 LEDs at a time. I don't know how much current you are trying to get them to take, but the Insructables page suggest 10 or 20 mA. Only one layer can be lit at a time, so you need quite a lot of current to make the LEDs show up.

At room temperature, the internal resistance of an AA battery is around 0.15 Ohms (https://data.energizer.com/pdfs/batteryir.pdf). You have three in series, so say 0.45 Ohms in total. Taking 640 mA (64 lots of 10 mA) will cause a voltage drop of around 0.3 V. That alone doesn't seem too bad, but with new batteries starting at about 1.6 V, you will get less than 4.5 V when the lights are on.

I think that the circuit is designed for 5 V. It uses a 74ACT138 which is only specified down to 4.5 V

It comes down to the fact that you are using batteries that are too small for the job.

I suggest getting a USB power bank and using that. If you have set the resistors to give 20 mA per LED, make sure you have a power bank rated at 2 A.

 

[benedpd]

Hi Diver300, thanks for your detailed response.

I apologise, I should have mentioned that I have powered this from my lab bench power supply (outputs 5.1V with well beyond enough current capabilities), and the result is the same.

I have my resistor values so the LEDs should be taking approximately 14mA (100ohm). I just tested with my multimeter and when a full plane is lit (64 LEDs) I measured 0.1-0.12A of current draw.

The fact that it works with my power supply is leading me to believe the issue is somewhere else. Any other ideas would be greatly appreciated.

 

[Diver300]

I can't get your videos to work.

The fact that it stops working when there are lots of lights illuminated makes me think that it's a power supply problem.

If each plane is 64 LEDs, and each LED can take 14 mA, then that is about 1 A. Your reading of 0.1 to 0.12 A is the average, but a steady 0.12 A is very different from pulses of 1 A for about 1/8th of the time. Your power supply may not react very quickly to the increase of current each time the LEDs turn on. It may be advantageous to have larger capacitors near the semiconductors.

 

[benedpd]

I can't get your videos to work.

I apologise, this has been a very bad showing from me. I appreciate your patience. I have updated the links with youtube links so you should be able to see them now.

If each plane is 64 LEDs, and each LED can take 14 mA, then that is about 1 A. Your reading of 0.1 to 0.12 A is the average, but a steady 0.12 A is very different from pulses of 1 A for about 1/8th of the time. Your power supply may not react very quickly to the increase of current each time the LEDs turn on. It may be advantageous to have larger capacitors near the semiconductors.

Yep that's a very fair point, I didn't think about the pulses. I have a couple of capacitors on the input power lines (1000u, 10u, 100n) and smaller ones on the output shift registers (100n). As can be seen in the pictures below. Would it be indicative of a capacitor problem if my cube worked for a month then now that I've come back to it 3 months later it isn't working? That is the part of this issue that is a little confusing to me. Maybe I somehow blew them.

By having larger capacitors near the semiconductors do you mean the transistor array below or something else?

Cheers

 

[Diver300]

I meant that you should have the capacitors near the ICs, rather than near the power supply. The inductance of the cables can cause problems if you are trying to change the current very quickly.

Electrolytic capacitors will blow if there is reverse polarity or there is an over-voltage, but they are a lot more difficult to blow in those conditions than the ICs. Electrolytic capacitors will degrade in time, but it takes years at room temperature for that to happen. It is more likely that some wire broke if the cube were moved, or there is dampness causing a partial short.