A Little Quirky Project Updated to V6

The board is 6" long when the top LED is cut off the board. The boards are designed to be stacked; 4 stacked is just over 1 foot if the top board is not cut. The LED spacing is 100_LEDS per Foot (plus 1).

LEDs are intentionally populated on the side without silkscreen.

The Driver_Board will be uploaded after it is properly uplgraded to V6. Hopefully someone finds this interesting.

Files:
04_LEDs-2023-02-06_192625.zip (357.4 KB)

Edit_1: The connector board with 2 boards stacked.

Edit_2: Alongside the Leveling Rod


In the photo, the LED board is upside down. When installed, it will be behind the leveling rod; with the leds along side the markings of the leveling rod. This constrains the height and width of the board. Also notice that by mounting the LEDs on the “wrong side” there will be no distracting white silk screen text visible.

Is it a stairway to heaven? :rofl:

Sorry, @Sprig, I just couldn’t help myself.

This is revenge on the installer)

1 Like

To me it seems more logical to make the split in between two LED’s instead of right under a LED. Maybe also put the connectors a little bit further from the edge, so single row bent headers fit. Or did you have other connectors in mind?

Some kind of collection of Open Sourced and PD KiCad projects would be nice. There is already Made with kicad | KiCad EDA (Which has gained a 6th page recently) but I think the threshold is a bit high for a lot of people.

Several PCB manufacturers also have viewable or downloadable PCB’s but often it’s only the PCB or the Gerbers and without schematic.

1 Like

I don’t know what you mean. The 3D model is a hack, the actual part number used for assembly is a 3 legged 2 color LED; center pin being GND.

There are several physical design restraints. The 2 most critical are that the boards must be able to source/sink 20mA of current for 102 LEDs and that the LEDs be mounted on an edge.

The reason for the location of the connectors is due to my idea of using a connector board in such a way that the boards can be “stacked”. I’ll get a photo of that uploaded sometime today.

Sorry, forget that. I had not looked closely enough and had not seen your intermediate PCB connection thing. Euhm, you probably added that later…
Well, it’s one way to do it, but at the price of an extra PCB, and the cheap fabs usually charge based on the amount of PCB’s. You could overlap the PCB’s and then compensate by the distance between the LED’s and the PCB, but that looks ulgy. I would probably just use edge connector like pads and then just solder wires directly over it, or use the square 0.1" headers, lay them flat on the PCB and solder them on both ends. Those steel pins are quite strong.

That is a point of concern. Connector placement is a simple factor that has an surprising big impact. Compare putting the power connector in the center of the PCB compared with at the end of it. When the connector is in the center, you halve the copper length (and thus resistance) to the end, but also halve the amount of current through the track, so that is a reduction of voltage drop by a factor of 4. the tracks passing the power between the PCB’s could also quite easily have been made much wider.

When opening the project, KiCad complains about a missing rescue library. It looks like you fixed the schematic symbols to use a “01_Devices” library (Which is missing) but just forgot to remove the rescue library from the project’s lib table.

There are lots of overlapping texts, such as:
image

The symbol of the 74HC595 is wrong. Pin 9 is not an inverting pin. In the datasheet it does not have an overbar but an apostrophe. It just bypasses the output latch but does not invert. (The standard symbol in KiCad’s library for V6 is correct). I also don’t like putting the QH’ signal in a row with the other outputs. An empty line here would have made it much clearer this pin has another function then that “group”.

In the schematic you have left the center pin open, but on the PCB they are connected to GND. It’s a bit confusing to define these symbols with a GND pin. Putting pin 1 in the center is also a bit unusual.

What is the size of these capacitors? 100n or 10u?
image

LED’s without series resistors?

Why put al these shift registers rotated on the schematic, and then combine that with the zig-zag bus? I would have put them just next to the LED’s in a straight line, even though that goes against the convention of maintaining a left to right signal flow. It looks like an lot of extra work to draw it in this way. I guess you redrew that section 3 or more times…
image

I’m also (unpleasantly) surprised by the PCB itself. The routing looks quite haphazardly without a decent GND plane, and putting the shift registers on both sides of the PCB is a bit of a nuisance for manufacturing.

Why put those buffer capacitors in the far corners of the PCB? They look mechanically quite fragile. They are also by far the highest parts on the PCB. I would have put them further from the edge, kept their legs a bit longer and laid them flat on the PCB (could have added some glue too).

I would have turned the PCB 90 degrees. That fits much better with a my landscape format monitor, and thus results in less zooming during PCB design.

But in the end, who am I to complain? It’s your project and if it works for you then that is the main thing, and you are of course free to ignore anything and/or all I have written. Quite a quirky project indeed.

1 Like

This was a conversion from V5 to V6, and in the process quite a lot of things changed (NOT everything for the better). Not every error was found.

Yea, I know that, I did not care to fix it when I when I was first working on the project. You might be happy to know that I have already made the fix to make it comply with the Datasheet.

At the moment I’m a little uncertain, but I thought I remembered that convention for pin 1 of diodes was GND; blame the manufacture of the diode for the strangeness.

Not sure… There seems to be something in the conversion steps that I overlooked. Thank you very much for pointing this out.

The intent was to try to keep everything on letter size paper with very little signals crossing over. If this were something I was being paid to create, I’d probably have tried to be a little more standard.

That is mostly due to an attempt at making the code easier to write. Did I succeed? I think so. On the board layout, the top layer top shift register returns the serial data to the bottom layer bottom shift register; and that makes a mess out of the PCB layout.

Both top and bottom layers have a ground fill that is unbroken. The board functions without issue at the rated speeds of the parts.

They are bulk capacitors to account for massive current draw with some future imagined illumination patterns. One lead had a cold solder joint and the test patterns used experienced some irregularities until reflowed.

This board is going to physically interface with 2 other physical objects that will be vertically mounted along side of it. I’m not sure I have the mental capacity to keep flip-flopping all 3 parts in my mind.

There was zero doubt in my mind that anyone taking a serious look at this first piece would at some point wonder, “Why did he do THAT?!” The simple answer is that there are significant physical constraints.

In my opinion, they are no longer needed with modern regulators and modern diodes that are appropriately matched. In fact, with this design, series resistors would have created significant headache and cost to the project.

@paulvdh Thank you for the time you spent to create the feedback you provided.

The 74HC595 has a maximum allowed current through VCC or GND of 70mA, so if all eight LEDs are on at the same time, the current per LED / output must be limited to 8mA or less. The output can drive up to 35mA (per single output), so there needs to be a current limiting resistor in series with each LED.
Perhaps that was what you discovered in the other thread with the picture of the dirty boards…

This topic was automatically closed 90 days after the last reply. New replies are no longer allowed.