5x5x5 LED cube help

The first 3 layers are shown. Columns are the anodes.
Each layer is made of 5 groups of 5 common cathodes - one group of one layer is turned on at a time.

CrossRoads:
The first 3 layers are shown. Columns are the anodes.
Each layer is made of 5 groups of 5 common cathodes - one group of one layer is turned on at a time.

but wouldnt turning on group 1 section 1 turn on that whole section? not individual LEDs?

I think I was imagining something different than that, but that is pretty interesting too.

It looks to me like that does all rows at the same time, but does one section at a time, meaning a 20% duty cycle.

I was thinking something that would do one section on one row, then sequence through each section, until it gets to the last section, then shifts to the second rows first section, sequences through each of that rows sections, and moves on to the next row, until it completes each section in each row one time. I think that would be a 4% duty cycle.

Actually building one of those seems quite challenging, but I could be wrong.

suicidalacorn, I think that anode pin controls the individual led, in each section of 5 LEDs, if I understand your question.

"wouldnt turning on group 1 section 1 turn on that whole section? not individual LEDs?"
Only if the anodes were driven high for all LEDs. Otherwise, just the LEDs with High anodes turn on.

The idea is you enable 1 group of 5 LEDs at a time, a 5x5x5 cube would have 25 groups to cycle thru. 4% duty cycle.
Driving each group for 1667microseconds would yield a 24 Hz refresh rate - leaving over 25,000 cyckes in between writes to do other stuff.
I would imagine a loop within a loop to cycle thru the groups & layers, reading from a 25-byte array (upper 3 bits ignored) to send out a new byte every 1667uS. 2nd array of 5 keeps track of group enable pins, 3rd array of 5 keeps track of layer enable pins.
There are 5 anode drive pins, 5 group select pins, 5 layer select pinsl, so the whole cube is multiplexe with just 15 pins, directly controllable by a single '328P chip with no extra shift registers. Just 15 resistors and 20 transistors.
I didn't make one myself, just helped the original requestor with a design that could be done.

Cool, thanks for the explanation, I had missed the third group of pins, now i see 3 and understand that the other 2 would go with the last 2 rows.

CrossRoads:
"wouldnt turning on group 1 section 1 turn on that whole section? not individual LEDs?"
Only if the anodes were driven high for all LEDs. Otherwise, just the LEDs with High anodes turn on.

The idea is you enable 1 group of 5 LEDs at a time, a 5x5x5 cube would have 25 groups to cycle thru. 4% duty cycle.
Driving each group for 1667microseconds would yield a 24 Hz refresh rate - leaving over 25,000 cyckes in between writes to do other stuff.
I would imagine a loop within a loop to cycle thru the groups & layers, reading from a 25-byte array (upper 3 bits ignored) to send out a new byte every 1667uS. 2nd array of 5 keeps track of group enable pins, 3rd array of 5 keeps track of layer enable pins.
There are 5 anode drive pins, 5 group select pins, 5 layer select pinsl, so the whole cube is multiplexe with just 15 pins, directly controllable by a single '328P chip with no extra shift registers. Just 15 resistors and 20 transistors.
I didn't make one myself, just helped the original requestor with a design that could be done.

oh ok. that makes more sense. i thought you were saying to have constant current driving the anodes at all times. so you would just have transistors to control the anodes (or an LED driver like you posted earlier)?

ok, so for driving the LEDs i plan on doing what is described in this video (or something similiar with the ideas recommended by you guys). this is what i was trying to describe before. but i will use a pnp transistor where he uses a MOSFET (because i had a hard time finding those MOSFETS and transistors are much cheaper). skip to about 32:30 in the video to see what im talking about. How To RGB 8x8x8 LED CUBE - THEORY - YouTube

You can drive the anode with 5 arduino pins. PNP transistors are not needed.

CrossRoads:
You can drive the anode with 5 arduino pins. PNP transistors are not needed.

i am now confused. wouldnt that draw too many amps from the arduino? if i have all the anodes of each layer connected and i have 25 LEDs per layer, that would be 500ma per pin from the arduino to power it. am i misunderstanding something?

You are misunderstanding - only 5 LEDs at a time will be on. That's why the discussion of cycling thru the 5 groups, and the 5 layers.
5 x 20mA = 100mA max.

CrossRoads:
You are misunderstanding - only 5 LEDs at a time will be on. That's why the discussion of cycling thru the 5 groups, and the 5 layers.
5 x 20mA = 100mA max.

oh ok. now i understand why you broke it up into sections. that makes sense now. so i wouldnt even need an external power supply for this since it will be using very little power. correct?

Correct - no external supply needed. Use superbright LEDs tho.

so does that mean i can use the pwm ports on the arduino? i do believe, if i remember correctly, the arduino uno has 5 pwm pins.
*edit: i looked it up and it has 6 pwn pins

Yes.
An Uno has 6 - 3, 5, 6, 9, 10, 11

awesome! thank you so much for all of your help! i really do appreciate it! ill try to post pictures of the finished cube when i finish it. i do believe i have all the information i need to feel comfortable ordering parts and not worrying about buying the wrong things.

I really want to build one of these, sadly I have a lot of things going on, so not sure when i will be able to get around to it. The difficulty, as I see it is the isolation of each row. I will be curious how you work that one out.

Thanks again crossroads, I may build one of these, but mostly I really like the idea. Nice work that.

After typing this post, I had a crazy idea. RGB LEDS. It would need a total of 25 control lines. I suppose you could do bi-color LEDs using a total of 20 control lines.

Yes, having 5 isolated groups per row would make it a little trickier mechanically vs having the whole layer connected.

RGB, yes add 10 more control lines.

Could use '1284 for that, 32 IO total. Could achieve same update rate, send out 2 bytes of anode info vs just 1.

you would be doing 300ma, instead of 100, You might not even have to use different transistors for the sourcing, so it sounds like you could do this without having to change anything else.

It seems to me that you could also do the anodes differently, like with shift registers or LED drivers. I wonder, if you serialized all the data, could you run this off an attiny?

That 32kb data limit may be an issue too, it seems that isnt much space for 4x4x4 RGB cube, and a 5x5x5 is twice the data.

I was thinking about the mechanical aspect, and Im thinking if we turn it on its side, so that the 25 freestanding iines of LEDs mounted to the bottom plane, (similar to the charliecube), it might work. I will have to ponder this a bit more.

I might add from experience building a 5x5x5 led cube to be sure to use diffused leds as the brighter leds are too directional in their light to give good observation from wider angles of view. Narrow spread leds will have a negative effect on a lot of the visual pattern effects, where diffused leds give better consistent views.

Lefty

I've read of people lightly sanding them to diffuse them, dipping white paint, putting a white ping pong ball over them ...