Oh and btw. The reversing of the input and relay numbers were simply pragmatic to the layout of traces on the board. I expected there to be a bit of reverse logic that will do my head in but I will code pins in an array so I only need to work it out once and the software will take care of it
Yes, with the emphasis on some.
Four of the LEDs linked to 'on' at the same time will drive the chip to the edge of it's thermal limits.
The chip itself could then become >100 degrees C.
Much better to use two or more outputs in parallel for >150mA loads.
Leo..
The problem is your current design has potential flaws that might lead to major issues.
When designing circuits we try to think of all potential problems that can appear under normal/abnormal conditions.
Yep, I was wondering about that and the potential to fail ON at the relays which is probably sub optimal. I will look at the other chip and not adding the resistors.
The uLN claims to be able to drain 2.5A which would be 10 LED âbulbsâ like the ones linked. Might be best to air on the side of caution anyway
Datasheets only look at the bright side. The user needs to read between the lines.
If you assume 1volt saturation, then 2.5A coul get chip temp to 25C + (2.5 * 70C) = 200C.
Leo..
Here is a mock-up of a switching module for those type of bulbs at ~250mA each. Using the PCA chip this time. I removed the resistors as these should not be required and thus I should be able to have 'normally off' LEDs. I have not put any LED indicators as this is for on a boat and everything is battery powered. I know they use very little but I wonder about their usefulness when the unit will be hidden away. I may eat my words!
Looks great. Ganging up the outputs is a great idea.
Cool logo!
Small suggestions:
- you could move C1 even closer to the PCA.
- the diagram still lists PCF8574 in the header in blue
Where does the 12V come onto your PCB ?
If your loads are powered by 12V, and pin 10 of the 2803s are connected to 12V, you can drive inductive loads within the current limits of the 2803.
You could use this PCB connected to a dumb relay board then use one of these modules to drive external LEDs and one of these modules to drive the dumb relay PCB.
You may want to consider a dual U4 header.
One column goes to the 2803 outputs, the pin beside goes to +12V.
This way a simple 2 conductor connector can be plugged onto the 2 header pins then going to the destination LED.
I will move c1. Donât know how it ended up down there! I think with chip changes etc it migrated. I caught the labelling in the schematic too. I was concentrating on the pcb so only noticed after I posted and was reviewing for purchasing.
Thanks
12v will actually come in on the bottom right mounting hole but only for the ULN. Common ground comes in at the top 2 mounting holes. I use things not entirely unlike this;
The LED bulbs will have a positive bus bar elsewhere which can be fused. I have adjusted the label to remove the non inductive element and instead assert a 2A total max which should stay well within the chips limitations. In reality it is unlikely all lights to be on at once but best to be sure.
I may go through many iterations but I find things like the headers/sockets to be a pain in the &@âŹ. There are not easily available and searchable lists of compatible ones and I end up spending ages going through the parts and ending up with random variations.
.
Any reason? They are just vias and pads. My thought was I could make a better connection with a standoff screw
Standoffs are fine but using a screw through a mounting hole to the chassis will become intermittent and/or oxidize.
BTW, if you use Plug-in PCB Screw Terminal Block Connectors it makes things easy to work with.
It is a standard method of electrical connection on a boat. All connectors are tinned and heat shrinked with glue to make watertight connection and the exposed terminal is sprayed with electrical lube. It makes a connection that canât pull out or readily fail with vibration.
I have seen those plug in terminals but canât find a simple source that I know will work. Too many options all thrown in a bucket together online.
Just offering you some suggestions 
.
That's Engineering 
.
EDIT
I have swapped out the PCF for a PCA as described and finally everything works. Had a few snags as the addressing DIP switches are back to front (in that they are marked on where I think off is more intuitive) and so I was addressing the chips as HHH instead of LLL for default I2C.
Of course my soldering iron gave up the ghost so I had to use a hot air gun designed for taking paint off to reflow the chips. This resulted in some melted relays until I built shielding around it! I also had to do testing with the oscilloscope to find the inevitable poor connections created by this, less than optimal approach. Also I only got a few of the fuse holders soldered in before it popped the breaker!
Now I will need to move to the coding stage of this project. Basic code turns the relays on/off as needed but I will need to build sophistication in. The control module is likely to have many additional functions so it will take some time but I will start posting code as soon as I get it up and running.
Really good to see things are slowly but steadily coming together!
lol. My process has not been optimal, I know. I will get something up and running then test then increment. The funny thing is that massive changes are likely but, as I learnt early in arduinoland, if you can dodge a wrench your can dodge a ball! Basically get it working any way you can and then start incrementing towards perfection. I want one of each of my modules working, then working together and then put them through their paces and improved as required