Tony's IoT Experiments

I was able to spend a bit more time on this project, and I got my local UI working on Sprinklers_pi code base.

The UI itself is inspired by the OpenSprinkler UI - home screen is identical and animation there is the same. However implementation is rather different, also Local UI functionality changed a lot - now you can use Local UI to control valves etc, in a way somewhat similar to the typical Orbit sprinkler timer. The UI can work with both 3 and 4 buttons.

Also I implemented logging sub-system for Sprinklers_pi on AVR/Arduino. It uses SD card for logs, stores information in CSV files (allowing easy download to Excel), but it is also compatible with the built-in Sprinklers_pi Log view.

The code seems to work (Alpha quality), but was not tested much. The code is available here.

Next steps: Logging sub-system is laying foundation for sensor data collection. I'm planning to add Waterflow, Temperature and Humidity sensors to my setup - this will be the next step.

Finally I was able to get back to work on my OpenSprinkler project.

Just to remind, I'm using a fork of OpenSprinkler 2.0 code, modified to work on the standard Arduino Mega 2560 with the mainstream Ethernet W5100 controller. I did some work/enhancements to that code before, adding SD-based logging and fixing some bugs.

Last weekend I also re-worked local UI in the OpenSprinkler code. The original Open Sprinkler uses fairly basic UI for the local LCD - it is mostly intended to show status and to configure few things (IP address etc). I wanted to add more functional local UI, comparable to the standard Orbit timer controllers - with the ability to start/stop valves, review and modify programs etc. However retrofitting this functionality into the existing OpenSprinkler UI code is rather hard - that code is optimized for space and is not extensible.

To fix it, I re-worked the local UI part of OpenSprinkler code, moving it into a separate class and making it extensible. The work is not fully complete yet, but I have basic local UI working and functional.

The code is available in my Git repository: https://github.com/tony-osp/OpenSprinkler-experiments

Next steps: I need to spend another weekend to finish up Local UI code to implement all features/capabilities I want (e.g. full local config, maybe program review etc) before I can move on to the next major update.

But the next major capability I would like to add is support for monitoring - water metering, temperature and humidity monitoring.

However before I start to work on monitoring, I need to decide whether to continue using existing OpenSprinkler code as the core/base, or maybe to move to the core code by Richard Zimmerman. Richard's code is a totally re-written version of OpenSprinkler designed to work on both Rasberry Pi and on Arduino Mega. Richard's code is much more comprehensive, it has more complete built-in WEB server and slick UI. Big limitation of that code is the lack of the local UI (no support for local LCD and buttons), however now I have this code and I can try to transplant my Local UI into Richard's OS version.

Last weekend I connected another group of 4 zones to the central timer using my voltage converter, and because there was no convenient spot to install the converter right next to the valves I had to use longer cable.

It turned out that Orbit 91592 valve works fine when connected via 45-feet long cable to the voltage converter box. The cable is a standard 18AWG 5-wire irrigation cable, nothing special. Also I found that 18AWG solid copper wires (as used in the sprinkler cable) are quite good for inserting directly into the Orbit valve custom connectors - I just put some electrical tape on the top and it is done.

Resistance of the 18AWG wire for 45 feet is about 0.3 Ohm each way (0.6 ohm total) which is not that high since the voltage I'm using to trigger the valve is 24 volt and there is healthy margin here. It will be interesting to try 100 feet cable, it should work.

This actually increases the appeal of using relays in the irrigation timer instead of thyristors, since these relays can directly drive Orbit DC valves (using the same simple circuit) and in many cases the wire length will be sufficient to directly connect valves to the central timer.

I successfully mounted all key blocks in a suitable box and have it running. Probably next weekend I will connect it to the sprinkler system and give it a try in a real setup.

My configuration/hardware looks like this:

1. Arduino Mega2560 R3 (sainsmart.com version)
2. Standard Ethernet shield with WizNet5100
3. Large size LCD 1602 (the same as standard 1602 but almost twice bigger - easy to read at a distance)
4. 4-channel Relay module
5. Power supply module (it takes 24VAC and produces adjustable stable DC)
6. Four buttons and power switch
7. Standard 24VAC brick



All this is mounted in a plastic box with clear cover. LCD module is mounted inside the box but is clearly visible through the transparent box cover. All modules are mounted on a piece of FR4 plastic installed inside the box.


Right now I have only one 4-channel relay module installed which allows me to connect 4 zones, however the design allows more channels and the box has space for 3 more relay modules (or a few bigger ones) - it should allow me to have 16 channels controlled by the timer directly.

All together it looks like this:




Some observations/learnings

To my big surprise WizNet 5100 tends to get rather hot when the box is closed. The chip is always quite warm, but in a box it gets rather hot. I ended up gluing small radiator to the chip to ensure it will be stable and reliable on a hot day.

Also I was surprised to see the large-screen 1602 module to draw quite a lot of power - 250 milliamp. When trying to connect it to Arduino Mega onboard power supply it gets pretty hot - this was one of the motivations for installing separate power supply module rather than using common 9V power brick. All this power is required for LCD back-light - spec sheet mentions 240milliamp power. I ended up installing 24VAC->5V DC power module, and I also added resistor to limit LCD back-light current to about 120 milliamp - it produces sufficiently bright back light but is easier on the power supply circuit, to ensure that power module stays cool.

Another big surprise was with control buttons - I installed standard pushbuttons (bought it from Jameco), but then I found out that these buttons are NC (normally closed)! This is quite unusual, I guess I did not check when ordering. Good thing is that I was able to work around it in the code, just changing the button detection logic.

There were also bugs in the firmware (the relay version, obtained here) - the relay control output code was confused with the way how OpenSprinkler handles multiple boards (even inactive extension boards), and with the order of bits vs order of stations. But it was easy to fix.

I was able to run the system for a few days with a basic program configured, and I can see in the log that zones were correctly turned On and Off as programmed.

Next Steps

I'm planning to install this sprinkler timer (together with my 4-channel voltage converter box) to run a section of my irrigation system, to get a feel of how well it works in a real life.

Also I'm planning to enhance firmware a bit - I want to add local (on the device) manual controls, to allow simple on-demand runs (e.g. "start zone 2 and run it for 10min) - this is quite useful for irrigation system tuning/maintenance.

Once this is done I will add more channels to the irrigation timer and install it for daily use.

Future Enhancements/Projects Ideas

I'm planning to put together few enhancements/new things to expand OpenSprinkler capabilities. I would like to add master/slave mode, to allow one OpenSprinkler timer to control several remote modules, connected via RS485 or other suitable link. Also I would like to add sensors (humidity sensor, temperature sensor, water flow counter) to the remote module to track the weather, soil condition and actual water use counters.

Another interesting option to try - I'm planning to experiment with Arduino module based on AtMega 1280p - a more powerful version of MCU, it has less IO than Mega, but RAM is 16KB - double of Mega. This will be helpful in adding all the extra features. Another option to consider is BeagleBone Black.

The Problem

Setting up comprehensive irrigation system is not easy and could be pretty expensive – you need to dig trenches and run water pipes, run irrigation cables and install control valves. It is not cheap even for a relatively short runs, but when you need to water sections of the property that are 200-300-400 feet away it could become really expensive.

I have a small-size regular irrigation system at my property that covers area near the house – 4 zones with a common timer, however this system is not enough. I wanted to expand the system to add more zones with drip irrigation and few sprinklers, but it would be fairly complicated and expensive – need to dig trenches, run pipes etc. As many others I went with a simplest option and just run few good quality garden hoses to connect drip irrigation sections and sprinklers, and setup Orbit battery-operated timer that connects to a regular garden hose.

The Orbit timer I used is a four-channel model – one timer controls 4 valves, all connected to a regular garden hose. This allows me to run 4 additional irrigation zones using one Orbit timer.

Over time my irrigation system grew in size, and I added few more sections with additional battery-operated Orbit timers, all connected with garden hoses.

This setup works surprisingly well – I’m using it for over 9 years now and it proved to be quite effective. Upfront costs are fairly low – 4 valves + timer + 4-way water splitter costs under $75 (in the local Home Depot), plus some money for good quality garden hoses, plus drip irrigation part and sprinklers. Running costs are also fairly low – every year I need to fix few cracked drip irrigation emitters, sometimes need to change a valve or two, plus hoses eventually wear out – although several of the hoses I bought 9 years ago are still working fine. All-up it turned out to be fairly good and cost-effective way to do watering without breaking the bank.

The only significant problem I noticed over the years is the inconvenience of setting and changing irrigation schedules – since the schedule is individually set on each 4-channel irrigation controller I must walk over to each controller to program it, I must ensure that the clock on each controller is set correctly, and that irrigation schedules don’t overlap – since running more than one zone (set of sprinklers) at a time will not work – not enough water flow.

Also I cannot really change my irrigation schedule depending on weather – it will be too cumbersome to adjust the watering schedule across multiple timers, so usually I don’t bother changing it much – which leads to overwatering and under-watering.

 

Central Timer

While walking around programming my irrigation controllers I was thinking – wouldn’t it be great if I can program all schedules centrally, using common irrigation timer?

I already have regular multi-channel watering timer (the one that drives 4 zones near the house), it will be great if I can connect all my irrigation valves to it and just program it all from one place. It would also allow me to attach rain sensor and/or use other advanced options for automatically adjusting watering times based on weather.

This is essentially what you get with a regular fully wired irrigation system – all valves are controlled from the central timer, lots of advanced options and flexibility etc. However I don’t want to spend big on digging trenches and installing full-blown irrigation system.

 

Voltage Converter Option

In theory it is possible to connect all my irrigation setup to a central timer if I install blocks of regular 24VAC control valves but connect them to a regular irrigation hoses. I looked into this option, but it becomes complicated and cumbersome – regular 24VAC valves are designed for plastic ¾” or 1” pipe connectors, and I will have to install size and thread converters for every connection. Also regular 24VAC valves are designed for underground installation and are relatively big/tall, so I will have to dig holes to install blocks of valves underground to hide them out of sight. Plus I will have to buy all new valves – while I already have Orbit battery-operated valves installed and running.

Another option is to somehow connect these Orbit 91592 valves to a regular 24VAC irrigation timer – in this case I can use all the existing valves and water distribution system I have in place, and can just run irrigation cable from the central timer to these valves. This is the option I ended up implementing.

Orbit 91592 Valve

The Orbit 91592 Valve used with these battery-operated timers is a relatively simple thing – it has two wires and it is operated by DC pulse. This is a latching type of valve – voltage pulse opens it, and opposite polarity pulse closes the valve. Unlike regular 24VAC irrigation valve no current is required to keep the valve in open position.

I found some information on voltage and DC pulse characteristics for this valve on the internet, I can recommend this excellent blog: http://rayshobby.blogspot.com/2010/06/minty-water-valve-controller.html

Essentially the valve is operated by discharging 2200uF capacitor, charged to 24V. Positive pulse opens the valve, negative pulse closes the valve. Ray (in the article linked above) wanted to control this valve using Arduino MCU running off a battery, and he put together a special circuit for it with step-up voltage converter to get 24V while running on battery, H-bridge etc. My problem is a lot simpler and somewhat different, since I would like to connect this valve to a regular 24VAC irrigation timer that supplies 24VAC to each channel when it wants to open the valve, and removes the voltage to close the valve.

After thinking through it I came up with a simple “low tech” way of doing it – using relay.

The relay is an AC 24V relay – it is directly connected to the irrigation timer. When the relay is triggered, the valve is connected to the 24V DC via capacitor C1, this generates current pulse of charging the capacitor, and this pulse opens the valve. While the relay is energized the capacitor is connected to 24V DC via the irrigation valve and is kept charged. When the relay is turned off, the same capacitor is connected to the ground and discharges through the irrigation valve creating negative voltage pulse, closing the valve.

This circuit is amazingly simple, but effective. It is cheap, easy to put together, and it is also inherently safe with regards to turning off water if power is lost or the irrigation timer fails.

This is actually pretty important – I need to make sure that the water will not be left running if power is lost or something else happens with the control wire/timer. Regular 24VAC irrigation valves don’t have this problem because as soon as the control current disappears the valve closes, but Orbit DC valves are of a latching type and they will stay open until the closing pulse is delivered. The circuit above solves this problem by automatically generating this closing DC pulse as soon as control voltage is removed, mimicking behavior of a regular 24VAC irrigation valve.

The 24V DC power required for this circuit can be readily obtained from the 24VAC available at the irrigation timer using simple rectifier. I also added voltage step-down circuit to limit the voltage to 24V.

R1 helps to limit maximum current via LM7824, also it helps to make the current pulse wider. Because current goes through the system only for a brief moment when capacitors are charged and the rest of the time current is zero, no heat sink is required.

I assembled a four-channel version of this circuit in a suitable weatherproof box installed near the irrigation valves. Because on the valve side opening/closing current can be quite substantial (although short), it is preferable to limit the wire length between the voltage converter and the valve. Because of that I mounted the voltage converter closer to the valve rather than on the timer side of the cable. Wire length of 15 feet between the converter and the valve seems to work fine, but I would not want to increase it too much.

The cable between the timer and the converter can be as long as required, the system is actually pretty tolerant to the voltage drop and long cable on that leg because 24VAC relays consume less current than a regular irrigation valve and are tolerant to the long cable resistance. Actual irrigation valves in this design are running off stabilized 24V voltage (and mostly off capacitors), and there is no problem with long cable run from the power source (24VAC).

I’m planning to assemble few more of these converter boxes to replace battery operated timers I have.