Looking for feedback on my first PCB design for my sump pump monitor. I plan to feed the board with +5V DC and use Veris Hawkeye 300 CTs to monitor motor status, and a stainless steel dual float assembly to track water in the sump well.
The relay on the board is to track the power supply status. I had mixed results using the boron's onboard charger reporting on a project deployed 3 or 4 years ago and the relay method has worked well for me.
Pump runtime and cycle count data will be fed via webhook to Losant once an hour. Power loss and water level alarms will be sent as they are raised/cleared.
The PCB seems fairly straight forward - to the point where I feel like I am missing something obvious. I plan to install headers so the boron could be swapped with an Argon if the site has satisfactory WiFi.
Any feedback is appreciated, even if it is "Your out of your league son".
Note from my side (this might be preference) but seeing that you are using THPs I assume it will be assembled by hand. IF this is indeed the case, why not increase trace widths. As you have ample space, it would not hurt and would make the traces less prone to possible damage caused by hand soldering.
One more thing, I would refrain from running traces that close to other pins. Again, it as you have a ton of available space, maybe rather route them out perpendicular to the pin and then change direction.
Lastly, I would use a GND plane on the Bottom layer and route on top as much as possible
With this is mind, there are some alternatives to the current transformer you are looking into. I am not familiar with the one you mention, not do I know whether there might be a specific reason for choosing that one, but a Quick Look on DigiKey (or Mouser and so on) and you will find ones much cheaper
I have no problem with 0.25mm traces, please do not get me wrong. It's just if you have space and you place to hand assemble/rework, I would suggest increasing the width. I am used to routing with 0.15mm but it is on boards with higher density.
Probably the one thing I would string suggest yes. Also, see if KiCAD allows you to add teardrops to your trace/pad connections. Not really needed here, but a good habit to get into (although heavily debated).
If you plan to use a cellular device, I would also suggest not routing underneath the module/modem. I have not used the dev boards before on PCBs and am used to working with BSoM units, but personally I would still treat those areas as Keepout Zones.
One more question; are you expecting higher currents through the relay?
Hope this helps and congratulations on your first PCB!!
I’ve had really good luck with those Veris CTs, I buy excess new stock off eBay for about $15 each. I have them on refrigerators, furnaces, freezers and of course, sump pumps. I have only had one bad one (knock on wood).
I updated the board as per your suggestions and it is at the manufacturer now.
It will be nice having a standardized board. Somehow I switched designs between the two prototypes I created and my semi-production version is wired opposite of my test version and I keep forgetting to flip the logic in software when I push updates.
Since most of the sump pumps that I've seen are Centrifugal Pumps, you might want to report AMPS also since you have the CT's. That's a good way to determine issues with the System, and detect any change in pumping performance over time.
AMPS will decrease if a clog develops in the suction or discharge side (reduction of flow).
AMPS will increase if the Forcemain/discharge pipe breaks (increase in flow due to reduced System Head).
I've also used AMPS to determine overflow conditions when the sump was surcharged well beyond the highest installed float (higher suction head = increased flow).
I looked up the Hawkeye 300 and it seems to be a switched device, not a true current transformer so as @Rftop say above, it would be good to know the current but I do not think that is possible.
I would add more input protection to the processor inputs. Perhaps a 100ohm serial resistor and two 3.6V zener diodes to the power rails. I might also add a series inductor for noise suppression but might populate a wire or zero-ohm resistor until I knew that I needed an inductor or ferrite.
It has been a few years since I tried this with an 4-20 mA CT. As I recall, I had a hard time getting accurate readings since the pump run cycle is less than 10 seconds (i.e. 26 cycles for a total runtime of 4.2 minutes).
I was using a Moxa PLC back then and by the time the readings got to my controller there weren't enough data points to make a reliable determination of true current draw/pump runtime. That system was really designed to monitor HVAC motors that had long run cycles. That was probably almost 10 years ago.
A quick google search indicates I will most likely need an external ADC to make this accurate. I'll keep searching and see what I can come up with. Might look at using the 0-5 VDC CT as that will prevent me from having to deploy a 24 V power source.
@bko do you have a schematic of the power protection you could share with me?
