MaddyP's 12G Nano Build - Node-RED Controlled

[MaddyP]

Hi All,

 

New to the forum but not to the hobby.  Wanted to share my nano build as an introduction.  🙂  This is not your regular build as most of the components were selected/designed specifically for this setup.  Hopefully you enjoy and glean something from it!

 

Current FTS:

 

 

Each part of the build can be involved so I'll post those separately.  More to come soon!

[MaddyP]

Design:

 

Ultimately this tank was meant to be a peninsula so the design reflects that pursuit.  I wanted to build out a tank with as little (or no) visible plumbing as possible.  Several designs were considered as follows:

 

Closed loop with plumbing to be hidden by rock rubble.  This would allow quick disconnects for components like Pax Bellum algae reactor, CO2 reactor, and probes.  The problem though was I wouldn't be able to run a skimmer in the traditional sense.  Plus the plumbing would be quite difficult to hide.

 

 

 

So another thought was to create a closed loop with inline sealed sump.  At this point I was really just spit-balling.

 

 

With another spin on plumbing.  I like this design as it would allow for a well sized refugium, but the sump would need to be custom and equipment would be limited.

 

 

At the end of the day (actually year) I settled on a much more complicated design which would allow a "traditional" sump and use of traditional equipment.  This plumbing configuration relies on modulating actuators to balance the water level in the display.  Pro: no overflow box or visible plumbing in display, allows for standard sump and equipment, cool looking!  Con: overly complicated, failsafe protocol necessary to keep display from draining, DIY controller.

 

 

And to finish off the goal of no visible plumbing in display, a false bottom to direct flow in and out of the tank:

 

 

 

This'll be a fun ride!  Proof of concept up next.

[MaddyP]

Proof of concept:

 

Since this idea of balancing water in the display hasn't been done before (that I know of) outside industrial automation, I wanted to confirm the concept would work.  I setup a test tank using a couple ikea storage containers and a quick Node-RED instance running on a Raspberry Pi. 

 

 

The interface boards are from Widgetlords Electronics for 4-20ma input/output.  

 

 

Modulating actuator is a Chinese branded AVA actuator I found on fleabay for testing purposes.  After taking it apart to confirm electronic components I'm convinced these are made at the same factory as the AVA actuators.  At some point I'll order a couple from AVA directly for peace of mind.  

 

  

 

And for sensing, a 4-20ma ultrasonic sensor sourced from Automation Direct.  Sensor range is 40-300mm (1.57"-11.81") with an analog output of 750-3750. This means ~11.5 points equates to 1mm of water level change.  

 

https://www.automationdirect.com/adc/shopping/catalog/sensors_-z-_encoders/ultrasonic_proximity_sensors/diffuse_300mm_sensing_range/analog_output/uk6a-d2-0e

 

I set up the concept using Node-RED for controll and ran it for several days with intermittent changes to the return power.  It held stable once tuned in properly!

 

 

  

 

 

This was enough to prove the idea would work, though it would take some fancy programming.

[MaddyP]

Lighting:

 

Every successful reef tank needs decent lighting.  I wanted to go sleek and controllable on this setup so naturally Nanobox Reef was top of my list.  I reached out to Dave to see if he would be willing to build a 6 puck fixture for my 12G long.  Two Duo fixtures would likely have worked fine, though the added two pucks would reduce shadowing to a large degree.  Luckily, he had a used fixture (old version) he would sell me!  

 

 

 

Originally, the plan was to run it with 0-10V from either an Apex/GHL so Dave outfitted the power supply with a PWM to 0-10V converter.

 

  

 

I really liked how compact these buckpuck boards were but because of the complexity of my design I needed to integrate an ultrasonic sensor into the LED housing to monitor display tank levels.  The sensor I chose was just slightly too large to fit in the housing Dave sent.  I used the existing Nanobox V3.1 pucks in a MakersLED Slim heatsink which was just large enough to accommodate the sensor.  Plus, I could fit the buckpucks directly into the housing to reduce wires in/out of the fixture.

 

After 6 pucks, 2 MakersLED 5up's, 1 ultrasonic sensor, and some Scrubs it came together. 

 

 

I chose to add a couple multiwire ports on one end for power, ultrasonic sensor communication, and 0-10V dimming for the 5 channel dimming pots (other end).  Overall I'm pleased with how clean the fixture turned out.

 

 

 

[DevilDuck]

This is a really cool idea! Can you elaborate on the failsafe for the actuators? It looks like your measuring the water level of the sump and not the display, how will you account for evaporation?

[MaddyP]

Controller:

 

Probably the most fun aspect of the build...and the most nerve wracking.  Control had to be DIY since the core function, a PID loop, would be keeping the water level consistent.  None of the commercially available controllers offer this feature.  At first I was going to try a PLC controller used in industrial automation, however ladder logic is beyond my current knowledge and SCADA design leaves much to be desired.  It didn't stop me from trying though.  

 

I ordered a Productivity 1000 setup from Automation Direct and interfaced it via Modbus to a Raspberry Pi 3.  It was a pain.

 

 

After some time I gave up trying to interface the two and settled on a middle ground.  Widgetlords designed their interface boards to work on Node-RED, a programming tool for wiring together hardware devices, APIs and online services.  It offers a robust foundation for these kind of projects with widespread community support.  And it's open source, meaning if I couldn't find a solution built by someone else I could build it myself in a language I already knew: Javascript.

 

Widgetlords offers a few different product lines with similar features.  I chose the SPI-DIN series since I wanted to mount it all on in series on DIN rails.  Although it isn't the smallest controller, there is a lot of control packed into this package. It is also expandable down the line by simply plugging in new boards. As it stand it's about 16" long, 6" wide, and 3" tall including DIN rail.  Features are battery backed real time clock, RS485, (2) digital inputs, (6) relay outputs, (6) analog outputs,  and (18) analog inputs. All with supported Node-RED nodes. Power comes through the main board (24V) and runs through all sub-boards.

 

 

These accommodate my need for analog and digital control, but don't offer hardware for probes.  Atlas Scientific was my choice to support probe readings since they are well known for modular probe circuitry and offer a clean interface for Raspberry Pi's via the Whitebox T3 carrier board.  These carrier boards also stack allowing for multiple probes if needed.  I went with two boards to support temp (x2), ph, orp, and conductivity.  

 

 

My goal was to have as many components on this build to run 24VDC as possible, but also needed to support equipment running 110VAC.  For that I chose the Kasa HS300 and Kasa KP303 power strips as they can be wirelessly controlled by the node-red-contrib-tplink node in Node-RED.  As an added bonus the HS300 has built in power monitoring to ensure equipment is working as expected.  

 

 

All of this (except HS300) is powered by a Meanwell HLG-480H-48 power supply rated at 48VDC 10amps.  This was chosen primarily to provide power to the LED with enough overhead to also run all 24VDC equipment via a Meanwell 48VDC to 24VDC converter.  

 

I'm sure someone will ask so here is the price breakdown on the controller components:

 

(1) Raspberry Pi 4 4GB - $55

(2) Whitebox T3 - $210

(1) Ezo PH - $42

(1) Ezo ORP - $42

(1) Ezo EC - $62

(2) Ezo Temp - $60

(1) PI-SPI-DIN-2x4MIO - $127

(2) PI-SPI-DIN-8AI - $66

(1) PI-SPI-DIN-4KO - $33

(1) PI-SPI-DIN-4AO - $37

(2) SDAFE - $72

(1) HS300 - $70

Total Cost - $876

 

This obviously doesn't include probes, wiring components, or misc stuff like wire raceway.  

 

Once the hardware was figured out, the next part was ensuring everything I wanted could be controlled by Node-RED.  It wasn't all straight forward in regards to reading the probes over I2C, but like I said before it's an open source platform.  So I built a node to help read the Atlas Scientific ezo modules:

 

https://flows.nodered.org/node/node-red-contrib-ezo

 

I also built out a set of UI nodes for use in Node-RED aquarium controllers.

 

https://flows.nodered.org/node/node-red-contrib-ui-reef

 

And the UI of my controller:

 

[MaddyP]

22 minutes ago, DevilDuck said:

This is a really cool idea! Can you elaborate on the failsafe for the actuators? It looks like your measuring the water level of the sump and not the display, how will you account for evaporation?

Thanks @DevilDuck!  There are a few redundancies on the system to be safe.  I have a ultrasonic sensor built into the LED housing to measure display tank level.  The PID loop runs off this sensor to ensure the display stays consistent.  The second ultrasonic sensor is measuring sump level and really is there as a backup, though it does add safety to my skimmer and helps with evaporation.  As for the drain line actuator, it has a built in failsafe which stores enough energy internally to close the valve in case of power failure.  It will also close the valve if it senses a lose of control signal from the controller at which point the display tank will rise to maximum causing the pump to shutoff.

 

Topoff isn't yet setup as I'm waiting for a 24VDC pump from Tunze to be delivered.  But once I do receive it I'll be relying on the sump sensor averaged over time to determine the water level for topoff.  The PID loop keeps the water level consistent enough (~1/4") for the ATO to be setup like most tanks.  

[MaddyP]

Final Setup:

 

And so with all the previous thoughts it was time to build it all and hope for the best.

 

Extruded aluminum stand built purchased from Misumi.  It's 40" tall with a ~30" sump space and ~8" on the bottom to accommodate ato / auto water change containers.  1/2" white acrylic was used for the surfaces.

 

  

 

Sump was built with a 20G long and Jax Racks baffle kit.  I ended up pulling out the filter sock component of the baffle kit due to excessive noise.

 

 

Plumbing is mostly straight forward with actuated ball valves in place.  Single drain with failsafe enabled actuator, single return split with three-way valve.  The three-way valve can direct flow to either end of the tank or both ends simultaneously.  

 

 

Electrical panel installed on the left.

 

 

 

Phoenix connectors for the win...

 

 

48VDC power supply mounted underneath.  This has an IP67 rating in case of spills.  I've also slightly leveled the tank tilting away from the electrical panel so if the display does overflow it will end up on the opposite side of the electronics.  And yes, I tested this by accident and it works...ended up with a gallon of water on the floor due to a programming error on my part.

 

 

Narrow wire routing to keep things clean.

 

 

And all buttoned up!

 

 

Finally wet!

 

 

And a partially sunlit moment...

 

[Maas101]

A cool feat of engineering and completely bonkers.. What's not to like 😀. My concerns would be the obvious mechanical / electrical failure leading to catastrophe but also the lack of directed flow without a powerhead and the lack of surface skimming ability leading to an oily film on the surface.

[MaddyP]

7 hours ago, Maas101 said:

A cool feat of engineering and completely bonkers.. What's not to like 😀. My concerns would be the obvious mechanical / electrical failure leading to catastrophe but also the lack of directed flow without a powerhead and the lack of surface skimming ability leading to an oily film on the surface.

Thanks! Of course mechanical failure can happen with any equipment. The actuators are setup with a feedback loop for reporting position and have internal alarm in case there is an issue. In all cases the actuator will close and I’ll receive a text alerting me to the problem. 
 

As for flow, each end of the tank has a semi-circle of acrylic with holes drilled at a 45 degree angle. This gives plenty of flow across the 35” tank from both ends with enough surface turbulence to keep scum to a minimum. It does build up a little but nothing that isn’t easy to mix in with a glass cleaning.

 

 

 

And one of the false bottom…