The next step was to finish off the diodes on their fly leads, covering the exposed leads with plenty of heatshrink.
I’m not at all convinced this is the best solution, but I’ve not come up with any better ideas where the prefab PCB is being used. Ideally I’d prefer to have the diodes fixed in position, much closer to the board.
I printed the lower section of the case, so it was time to fit the Nano to the PCB and put the PCB loosely in place so I could connect up the USB cable and see what happened. The good news is that it all lit up as it should and the LED that should flash did indeed do so.
I think it’s now time to start fitting all the bits and pieces that will go into the base (fan, USB socket, power socket, humidity sensor).
Temperature sensor plugs and sockets…
My dissatisfaction with the 6P4C connectors led me to try to find other options for plugging these in. Initially 3.5mm stereo jack plugs such as these sprang to mind.
However, Robert Brown himself explained to me that this idea had already been tried and had been found to have a couple of problems. First, the possibility of cross-connecting the inputs when inserting or removing a sensor connector with the power applied (which perhaps wouldn’t be a problem if this was always done with all power removed). Second, experience from people who have tried it shows that the currents involved apparently can lead to the plug and socket “welding” to each other because of the small contact area. This is clearly not desirable.
In the end I decided to try some 3-pin GX12 plugs and sockets (sometimes called “aviation connectors”):
In fact these don’t take up much more space than the original sockets and are much easier to fit and solder wires to.
I modified my design for the lid of the case and printed it out to try the sockets it. They seem to work well enough in terms of layout.
The same set again, but I had a particularly poor night’s sleep and really didn’t feel as though I had any energy. I managed ten reps before failing on the first set with one more fail before completing it, but again the 25s were very tough and I couldn’t even make ten.
Same set again today. This time I got to fifteen reps before my first fail in the 50m part of the set. I struggled with the 25s though, and couldn’t even complete ten.
I must post the full story of these events some time because it beggars belief, but after an engineer visited on Friday we no longer have a “whirly wheel” electricity meter. I think in fact that it may well be the first time in my life that I’ve lived somewhere that doesn’t have such a meter. I’ve kept it though, just for old time’s sake.
Instead we have a flashy new polyphase meter that could probably even play video games if there was somewhere to plug in a screen. Not a “not really smart” meter, mind. I’m told many of those won’t play nicely with solar PV. And although this one is supposed to support automatic meter reading it doesn’t do that because there’s no signal in the cellar where our electricity supply enters the house.
Same set again, and I did somewhat better than the previous time, failing on reps twelve and sixteen. I couldn’t manage many of the 25s though. My arms were just too tired.
For the final bit of PCB soldering, the MOSFETs and voltage regulator. The heatsinks I bought turned out to have small steel “legs” on when they arrived, which I didn’t initially realise was the case.
I assume they’re there so they can be soldered to the PCB and stabilise the heatsink. And perhaps to connect it to ground. There’s no place for them on the PCB though, so I snipped them off with the side cutters.
Because there’s not much room on the PCB for the heatsinks (some of the resistors prevent them from fitting flush with the PCB) I decided it was easier to put the whole thing together before soldering it into place.
Even then it was easier to put some of them into the board at a slight angle to get sufficient clearance. An alternative might be to just saw 3mm or 4mm off the bottom of each heatsink. I don’t think it would compromise their performance that much. And again I held them in place with a blob of blutak whilst soldering. And here they are, all done.
To make much more progress I probably need the case printing, so I’ll get that under way. I’ll also go over the back of the circuit board with a magnifying glass just to check there aren’t any unintentional bridges in the soldering. Then I’ll plug in the Nano and at least see if it lights up and runs the default “blink” code ok.
Fly leads for the diodes attached. No diodes as yet. I’ll cut the leads to length and fit the diodes once I’ve printed the case and can test everything for size/placement.
All the pin headers that I’m going to use are now in and there aren’t too many holes left, especially as the two four-pin headers that are left won’t be used (the ones for the Bluetooth connection and the HTU21D humidity sensor).
For the pin headers I used much the same method as for the resistors: put a number in place and then gently press a flattened piece of blutak down over the top to keep them all in position whilst the board is flipped over and everything soldered into place.
As I mentioned before, the five diodes have leads that appear to be 1.1mm in diameter and are too big for the holes in the PCB. In my previous build I soldered the leads to short lengths of wire that would go through the holes and fitted the diodes vertically, but that actually makes some of the jumpers awkward to reach with my fingers, so this time I’m going to put fly leads on the board and leave the actual diodes to sit beside the PCB in the case, perhaps fixed down with a bit of hot melt glue.
