Hobby Components 8x8x8 Cube Kit (HCKITS0050)

[Ramshackle]

Hi Andrew, have some progress, with all the grids out and the chips removed I made sure all the pins had continuity from the ic sockets to the pads on the rear side of the board, they all buzzed out ok, so reinserted the ic’s - and decided to check the grids again, only this time used my bench supply set to 2.5v with current limiting and check the leds again forwards and reversed bias.

I discovered a total of 7 more leds were either drawing 40 - 50ma forward biased in 4 of the grids and the same reversed biased - a couple were bright as the good ones but were drawing excess current, and one was dead short, (all the good ones draw around 4 - 5ma - changed all those out (I am now short of leds! ).

Put grids back in one by one and each time and watched for problems (powered the cube with 2amp iPad charger) boot sequence normal.

8 grids in and boots fine from the 5v iPad charges in the serial connector pins, demo sequence seems to run ok right from the beginning - oddly I tried to use the bench supply set to 9v - now it gets weird... the feet led light up but no led scrolling just a delay and the feet flash.. then a delay and what looks like a random display choice, unplugged bench supply and plugged in a wall 9v stabilised supply I use for my amateur audio gear... same result as the bench supply.

At this stage I decided to do some more checking tomorrow and see what the onboard regulator is pushing out.

Other than the above, minor issues seem to be straggler ghost leds either odd ones or a vertical line on the sine wave demo (never had those before)

Fresh pair of eye in the morning I guess.

[Ramshackle]

Hi Andrew, bit of a newsflash post, awake early today and decided to sit at the bench and try and look for clues at the erratic behaviour of the cube, with my limited knowledge, I watched what the cube was doing with each of the power supplies as follows.

1. iPad usb charger in the serial pins gnd and vcc, 5.2v and 2amp (original apple supplied)
Result: cube booted up and ran through demo reasonably well but seemed slow? stopping to pause and whenever it was picked up by the corners it would sometimes reset and start again, but always seemed like it was unstable.

2. Bench supply 9v plugged into rear socket.
Result: cube would light at the feet, but no boot up sequence (54321 etc) delay, then would enter into starfield demo, then skip, stop and sometimes reset and try again.

3. Plug in wall supply 9v 1.5amp charger from my icom amateur handheld radio.
Result: pretty much same result as 2. above, moving the cube would make it restart, not attempt to do anything other than light the feet.

The one common denominator was touching the cube grid or just having a finger near the atmega made the unit start and run, play at least one sequence and then stop.... it was like the atmega wasn’t running correctly?

I then checked all the pins again on the atmega for pads and continuity and all seemed ok. I did narrow it down to pins 15 and 16 (bottom right from the notch) just touching those and the cube sprang into life... crystal? Ceramic caps? It did seem like the crystal circuit, internal clock? Wasn’t working as it should.

Rightly or wrongly it seemed to me like a grounding issue?

Desperation made me think it was like those 2 pins were floating and needed to be tied down, I had a spare 104 cap and soldered it across those 2 pins..... the unit immediately sprang into life, booted, ran the 54321... then off it went.... ALL the power supplies work, wherever I plug them in around the house, moving it, picking it up and it works fine, doesn’t miss a beat, I even noticed a PAC MAN sequence I had never seen before, it’s now rock solid, and apart from one or two faint leds that appear now and then it’s working fine. (Maybe they are dodgey leds again, at the moment I’m just happy it’s working and maybe check them out later when I get some more leds. Can you supply a few of those?)

It’s been on the bench now for a few hours and runs perfect... I can turn it off and back on again.. 54321.. and it runs.

No doubt you will have a fair idea what’s going on here, broken track? Maybe the atmega itself is out of spec? Who knows, it’s above my basic knowledge to be sure!

Sorry again for the long post, but thought you would be interested and can maybe point to the offending component? Either way thanks for your patience.

[andrew]

Result: cube booted up and ran through demo reasonably well but seemed slow? stopping to pause and whenever it was picked up by the corners it would sometimes reset and start again, but always seemed like it was unstable.

cube would light at the feet, but no boot up sequence (54321 etc) delay, then would enter into starfield demo,

The one common denominator was touching the cube grid or just having a finger near the atmega made the unit start and run, play at least one sequence and then stop.

All the timing on the cube is ultimately controlled by the crystal circuit on the ATMega328 (U2). This is a tuned circuit and should (very accurately) only run at one speed. If the animations are running at different speeds then something may either be wrong with this circuit, or something else is effecting this circuit. It seems you have the cube working now but the things I would have suggested is to try and lift the crystal a little so it's not flat to the PCB just in case some rouge solder or flux is causing a problem and also check the two 22pF capacitors (C13 & C14) and their connections to pins 9 and 10 on IC U2. However, given the weird issue were the cube works better when you touch it, this could still point at a bad ground somewhere.

I did narrow it down to pins 15 and 16 (bottom right from the notch) just touching those and the cube sprang into life... crystal? Ceramic caps?

Desperation made me think it was like those 2 pins were floating and needed to be tied down, I had a spare 104 cap and soldered it across those 2 pins..... the unit immediately sprang into life, booted, ran the 54321... then off it went.... ALL

Sorry again for the long post, but thought you would be interested and can maybe point to the offending component?

Pins 15 and 16 are the 'enable' and 'latch' pins that control the 4 LED driver ICs. Adding the capacitors would help clean up any noise on these data lines but again noise on these lines could still be a symptom of a bad ground somewhere. It's the only thing I can think of (assuming there isn't more than one issue) that would explain all the things you are reporting. I don't know if you want to investigate further seeing and you've managed to get it working but the next thing I would have suggested is to remove U2 and check the continuity of the socket to the soldered pads for that IC - especially the ground pins.

Maybe they are dodgey leds again, at the moment I’m just happy it’s working and maybe check them out later when I get some more leds. Can you supply a few of those?

If you can email sales [at] hobbycomponents.com and reference this thread we can sort it out from there.

[Ramshackle]

Hi Andrew, many thanks for the reply, your diagnosis sounds better than mine! I will lift the xtal off the board slightly, check the caps around it again and report back, (will remove the cap I put in) and see if that brings any results, I may well put the cap back even if lifting the xtal solves it, as you say it’s not doing any harm.

Happy it’s working, even if I may have just masked a problem, may well just leave well alone I think.

Will also email regarding the leds.

Many thanks for all your time

[skyking]

Hi, newbie here.

First ever cube built, fingers firmly crossed and switch on. Works perfectly... well almost perfectly.

Demo mode runs as expected but the first four rows of leds on layer 3 (32 in all) don't light, at any stage.

Nothing obvious that I can see... any suggestions?

I have multi-meter, scope etc on hand.

Many thanks in advance.

[andrew]

Assuming I've understood the issue correctly - IC's U5,6,7,& 8 control each column so if any other LEDs on the same column are working (which in your case sounds like they are) then the problem can't be with that part of the circuit. IC U3 controls each layer so if any other LED on the same layer are working then again it can't be that part of the circuit either.

Therefore, it's likely the problem is with a bad connection on the cube itself. You need to look for a bad joint(s) somewhere around those four rows. In particular look where wire running from socket L3(?), the one which connects to that layer, and check all the joints between there to the faulty rows. If it isn't clear where I'm suggesting just post an image of the cube and i'll mark where to check.

[skyking]

Many thanks Andrew.

Had a look at all the joints from the point of connection on the +ve supply line on layer 3 and found the culprit - simple dry joint! All working now 100%.

Cheers...

[IBI]

I'm building this kit (v1.1) but I've got to a point where the instructions are unclear

The large LED instructions say "When placing the LED the shorter negative pin should be positioned to the flat edge on the PCB footprint."

LED.jpg

I understood the flat edge of the footprint to be the flattened part of the circle where the (obscured) D2 marking is but the pins are equidistant from this, so one can't be positioned nearer than the other. Have I misunderstood the instructions?

All four corners are have the flattened side of the circle equidistant from the pins, although top and bottom corners have it on opposite sides.

I see that this flattened circle markings on the diagrams in the instructions are oriented 90° from the ones on my PCB but as that seems to be the v1.0 kit I wasn't sure it was correct for v1.1.

[andrew]

Sorry about the confusion. It appears to be a silkscreen mistake that crept into V1.1. For correct orientation the short leg of the LED (cathode) should be soldered to the square pad on the footprint, and the long leg of the LED to the round pad on the footprint. This is the case for all 4 LEDs. Hope that makes sense.

[IBI]

Yeah, that makes sense.

On the subject of square pads and printing there is something else I'd like to double check.

The instructions for the electrolytic capacitors read "When placing the component the silver stripe on the capacitor should be on the black half of the footprint."

You can see from the picture in my previous post that would put the negative poles/silver stripe on the outside of the C7/C9 pair, through the holes with square pads.

However, the diagram in the instructions shows the square pads in the same place but the colours and capacitors the opposite way round to the text instructions.

pads.jpg

Capacitors.jpg

Are the text instructions the ones I should be following?