A SMT version

PCBWAY forwarded to me a query from a customer about the THT version of the board which is published as a shared project on their site. The person asked a BOM for the PCB. I explained that I didn't recommend the THT version for assembly as it used through hole components, but that I had already designed a SMT version but not produced it yet.

PCBWAY offered to sponsor the fabrication and assembly of the SMT version. I submitted the design from KiCad 8, using the PCBWAY external plugin which is very nice and generates the required files, then takes you directly to the ordering page on the website.

4 weeks later the boards arrived, extremely well packed. PCBs are actually quite robust but there were layers of padding in the box, just in case.

I connected one board up to my trusty Arduino Uno and loaded the test program sr32gpo-test.ino. Two LEDs are connected to the far end of the shift register chain. As expected they blinked alternately.

You can find all the necessary files at the Github site for sr32gpo-smt. Bear in mind that you may wish to modify a couple of things in the BOM. One is the value of the output resistors. I have chosen 4.7k which will source or sink about 1mA. The limit of the 74HC595 is 6mA per output. I also chose the LDO regulator for 5V, although it is not used in this test. You may wish to choose the 3.3V version if that's what your design needs.

Thanks to PCBWAY for the sponsorship so that I could validate my design. Now what do I make with these boards? 🤔

I've put another of this board to use to drive an alarm clock display. My bedside radio/alarm clock is failing so I'm replacing the dedicated clock chip from the 70s in it with a modern MCU based design. I have a second copy of the LED display used there with all segments brought out to the connector. The display is dim despite a current of about 5 mA per segment due to the era of the LEDs but that's ok for a bedside clock; you don't want to be kept awake by the light.

The display is being tested so it's upside down. But even if you flip the display mentally, you'll notice that the time is strange, it reads 80:32. The reason is the test program is sending out the serial data most significant digit first, so the correct time is actually 23:08. It's just a test program so I'm not bothered as long as it proves that the display is working.

I'll write up the alarm clock project when it's done.

Update, 2024-10-08: The write-up is at #Third life for a radio alarm clock

I have published a SMT version of the board, ready for PCB assembly to avoid hand soldering a whole heap of resistors and capacitors. It follows my investigations into the cost. You can find the link to the Github repository in the link section.

This version is not tested, as I currently do not have a need for more expanders, as I still have a lot of the THT version, used for example here. However the circuit design is unchanged, only the components have been rearranged on the board, and BoM and CPL files generated for a sample manufacturer, JLCPCB. So it should work, touch wood. I'm publishing it in case someone can use the design.