RS485 Mayfly Wingboard using open source hw

[neilh10]

Mayfly Wingboard using open source KiCad design with reliable components.

Board Requirements

• Compatible with Mayfly software . Mayfly connections

• https://github.com/EnviroDIY/Mayfly-Modbus-Wing/tree/master/knh002-MayflyWingShield

• Simple assembly/soldering to build

• Interface to two wires on Myafly (no direction control)

• good grounding option for potential lightening conductance Lightening protection suggestion EnviroDIY/SensorModbusMaster#15

• site install option for RS485 terminating 120Ohms

• one remote instrument may have 120Ohms

• efficient power connection, minimize chance of processor reset.

Optional

• current bleed ~ hardened hardware or software Add to Modbus an AltSoftSerial "flush" or "end" function, to set pins low EnviroDIY/ModularSensors#140
• Line Activity - LED Tx/Rx or combined. For non responding instrument there are three short Tx flashes as it polls. For responding RS485 there is a short Tx then Rx flash. How visible should this be in sunlight.

This is a continuation from
EnviroDIY#14

[neilh10]

I've got a basic board tested, and running with Mayfly

I need to refine the cct before discussing it - and what are the cost in build and $ of adding more options.
By removing the layered Amazon board, there is less space - but board could be extended.
Also can add more hw features do the hw bleed protection with an 74LV14 technology inverter, still investigating options. The 74LV technology has an Ioff circuit that turns off output and input when no power.
If LEDs for TX and RX to be added can use 74LV14 but again adds more components.

[neilh10]

Well I've got really behind in testing this.
The basic circuit uses a MAX22025AWA+ with decoupled RS485 Line driver from 5V, and 3v3 interface for Txd/Rxd
I use a variant to https://github.com/neilh10/AltSoftSerial that supports turning on/off the physical pins Txd/Rxd
In addition, for the boost this derives power directly from Vbat, using an idea similar to that of the "LTE Adapter" - and the boost is switched by the 3V3.

The final Mayfly test system interfaces to a Keller Nanolevel /RS485.

I've got it part of testing as part of a aggressive soak test, but what I thought was going to take 2hours to do, has taken a depressing 4 weeks, https://www.envirodiy.org/topic/stability-testing-how-to-do-it/ and is still on going.
The soak testing is working with an aggressive 2minutes sampling and posting over WiFi S6B.
The two minutes was to really test out the whole software stack/0.27.5 - from reading the sensors to posting to MMW.
The Mayfly power supply for the Xbee is marginal for WiFi S6B, and I did some testing with an improved external power supply, but it has made no difference. So probably a software stability issue has some how crept in.

initially I had a 2nd system with WiFi interfacing to Nanolevel/RS485 & LT500/SDI12 with 0.27.5 and these problems came up

neilh10/ModularSensors#21
EnviroDIY/ModularSensors#344
EnviroDIY/ModularSensors#346
EnviroDIY/ModularSensors#347

Well this is why all design, software and hardware, has to be organized towards what will be pre-release stability testing.

So now with the above solved on my fork, I've got it working with the target LTE modem.
Since this uses a data plan, I'm testing it as per what I expect it to be used in the field - 15minutes sensor readings of the RS485

So the test system using the KNH002r2 (KiCad board) is

and its ran from

2021-01-14 14:48 to 2021-01-19 10:58:00 - recording in excess of 3500 data points with no problems. So that counts as a good circuit verification

[neilh10]

I've put on a buffer SN74LVC2G34DCK, LEDs on the line Tx and Rx, and RS485 termination resistors.
Initial testing this works nicely. The Red on Tx andGreen Rx - considering baud is 9600 - when there is sucess the Green Led flashes brightly 3 times, and dominates the Red. When there is no response, the Red flashes brightly enough ~ 3 sets of 3 times. So the visual indication of working line is good.

The cct below "Rev4" is proposed, and put up for discussion . Any comments.

The objective is for the board to be hand solder able as per the initial RS485 wingboard; except for the specialist LTC2942 - which will can be done through a slightly more specialist SMD oven process.

Pulling the LED power from +5V has to change, and possibly the LED itself. The LED is a small footprint SMD, that requires careful assembly/placement - will look for an easier device to hand place.
The LTC2942 "Battery Fuel Cell" hasn't been tested - its an SMD DFN-6 part with a pad underneath - the initial solder paste hole was too large, so it floated on the solder paste and I need to retry the board.

[aufdenkampe]

@neilh10, this looks very interesting and promising!
Do you share the KiCAD files or Bill of Materials anywhere?

I'm wondering how to take the next steps for testing and/or manufacturing?

@GArrigotti-cws and I are in need of >30 Mayfly Modbus Wingboards by early spring (April 1?). How can we build off your designs to get us there?

Is the next step for you to send me one to test?
Or are we ready to find a way to manufacture these? @GArrigotti-cws and I could do the bulk of that work with your guidance.

[neilh10]

@aufdenkampe good to hear it could be of interest and timescales. I just published it yesterday - I have two PCBs that I can build.
This is to solicit any feedback before I do the next PCB turn and get more PCBs, and publish the hw files.
I can solicit the costs of separate assembly house to build 30 and see what the quote is.
I can split the hardware into a separate repo, or if enviroDIY creates a separate one I can fork that

[neilh10]

Do you have an assembly house you've used.
I have a couple, but are probably expensive.
One place that was badgering me last year is circuitHub
I'm exploring how they work - but looking at what they do, this maybe a similar example. For the cost they are proposing the production timeline, probably influenced by China New Year holidays, is Mar 22,
https://circuithub.com/projects/schell/fingerwell/revisions/9225/parts

[aufdenkampe]

@neilh10, it sounds like you want to do one more revision before scaling up manufacturing. I'm OK with you testing or for you to send me one if you would see a benefit in that. Either way, it might make sense to for us to connect via Zoom for a demo.

Can you share the link to the KiCAD and BOM files? I do not see any recent commits to this repo, or a link to where you might have shared them.

I don't have a preferred manufacturing house, but I could ask @s-hicks2 and @bschulz1701, or we could use that one you mention.

[GArrigotti-cws]

I’d definitely be interested, so I can submit analysis to my office.

[neilh10]

I have pushed above files for checking out - this doesn't include kicad yet as turns out the Rs485 cct lib parts have not been stored where I expected them to be.
There is a readme.md
All testing rev3 gone well with Keller Nanolevel on RS485 interfaces overnight.
I Plan on building two more rev3 circuits with fuel gage and verifying if it works before submitting next rev 4 for a build.

I need to know any feedback on circuit, what options are the most valuable.
What if any RS485 connectors are used on the board - there are three, either 4pin or screw terminal. I find I'm using a non standard arrangement 4 pin from the top
My thoughts are to get a quote with no RS485 connectors (post hand soldered), and also assume a link to the LiIon bat.
I'm Undecided as yet if it makes sense to add the FuelGauge to all boards. Its SMD and can't be added later.

[aufdenkampe]

@neilh10, thanks for uploading the rev6 schematic with commit 607cfe0.
https://github.com/neilh10/SensorModbusMaster/blob/release1/hardware/knh002-MayflyWingShield/rev6/knh002r6_rs485schematic.pdf

What are the differences with rev 4?

Could you also share the Gerber files, etc?

Last, I'm trying to understand the if there is a difference between the two JST power plugs. It looks like the "To Mayfly" JST has a small capacitor in line. I hadn't noticed this before and am wondering about the wiring. The pair of JST connectors on the Mayfly, the Sodaq GPRS Bee, and the LTEBee adapter are in parallel and functionally equivalent, but it doesn't look like these are. Can you describe your design?

My main concern & need is the ability to charge the battery from the solar panel or even the USB plug. In my test of rev 3, it doesn't look like the battery will charge if I plug it directly into your Modbus Wing.

On the other hand, I am able to plug my battery directly into the Mayfly, then wire the jumper from the other Mayfly JST to the Wing's "to Mayfly" JST, and everything works seems to work well.

[neilh10]

@aufdenkampe yes happy to release a full description, gerbers, BOM, its just on the slow track with everything else in the works.
Ideally the manufacturing would also be on a link and then add payment and it would be manufactured - but they aren't there yet.
The description of the board is here including new functions https://github.com/neilh10/SensorModbusMaster/tree/release1/hardware/knh002-MayflyWingShield
The board I sent as a rev3 was for proving the RS485 interface and powering to the instrument.
When used with the battery, it is meant to the battery plug in to the "Bat" connector and the Mayfly link plugin to the "May".
The benefit of the battery nominal LiIon 3.7 being supplied is to deliver the power directly to the boost U2 through U3 the power switch SIP3243.
The 2nd plug is a) not take up two plugs on the Mayfly as will need to have the power jumper to Celllular board. b) have a battery monitor - voltage and a coloumb/mAh counter in the link.
For rev 6 this is the 30mOhm R5. Happy to talk more.