[Ray]

To use the hosted app, you need to provide your router’s external IP address (or a dynamic DNS name if you’ve set up dynamic DNS for your router).

[Ray]

To Pablo and Dan,

Thanks for your suggestions. Since OpenSprinkler (microcontroller) and OSPi are based on different hardware and processors, it’s not always possible to keep the software features in sync. I agree that a comparison table would clarify a lot of confusions. But as the OSPi software is undergoing frequent changes (thanks to Dan’s continued contributions), it’s not always easy to keep the comparison table accurate and up to date.

I admit that I am lagging behind on feature development on the microcontroller version, but I am working on getting a new firmware released as soon as I can. The biggest upgrade in the new release is to enable the built-in SD card functionality on the 2.0 hardware, which will allow the controller to serve Javascripts locally, store program data on the SD card, and store logging data on the SD card as well. Some of these changes (e.g. serving files locally) have been implemented, and some are still in progress.

Another heads up is that I have finished the first prototype of OpenSprinkler Beagle (for BeagleBone Black). I don’t have much experience as to comparing Beagle with RPi, say in terms of reliability, but I know some users prefer Beagle for various reasons.

Speaking of embedded Linux, the Arduino team seems to be pushing in this direction. After the Arduino Yun, two new platforms have been announced: Galileo and TRE. Will definitely keep an eye on these new developments.

[Ray]

@tritron: note that there are two different versions of OpenSprinkler, one is based on the Arduino microccontroller and has buttons and LCD, the other is based on the Raspberry Pi (it’s called OpenSprinkler Pi, or OSPi for short), which does not have buttons and LCD.

[Ray]

I don’t think you need an external 24VAC to 5VDC converter: OpenSprinkler has a built-in switching regulator that does the 24VAC to 5VDC conversion. The 5V output is accessible from the VIN pin (in the pinout area).

[Ray]

@nhorvath wrote:

I developed my own relay board for the raspberry pi because I had 9 zones and didn’t want to have to use an opensprinklerpi + expansion just for the one extra zone. I posted the design on github in case anyone else wanted to use it: https://github.com/nhorvath/sprinklers_pi/tree/master/hardware/Sprinkler%20Controller

Nice design. Thanks for sharing. You may want to make this a separate post to get it more noticeable, or post it under the Hardware section.

[Ray]

28.6 should be fine. So I would suggest replacing IC1 first and replace the transformer later if necessary.

[Ray]

Looks like IC1 is damaged and needs to be replaced. Make sure your sprinkler transformer’s output is no more than 28V AC as that will likely to exceed the limit of IC1. Keep me updated if the issue has been fixed. Thanks.

[Ray]

Indeed the rain delay data is not recorded in EEPROM, so it doesn’t get preserved if a restart happens in the middle. This should be pretty easy to implement though.

[Ray]

There is, but it will most likely be semi-assembled, with some pre-soldered SMT components and the rest through-hole components for users to solder.

[Ray]

I think this is possible. OpenSprinkler uses Microchip’s ENC28J60 Ethernet controller, which is one of the lowest cost Ethernet controller. It does not have hardware Tcp/Ip stack, so the downside is that the stack has to be implemented in software. But with respect to IPv6, this is also the advantage: that is, it’s just a matter of changing the software to incorporate IPv6. I’ve seen a couple of online posts that discuss this. Google ‘ENC28J60 IPv6’ to see related posts.

[Ray]

I figured that it’s probably due to implementation differences in browsers. I tested both Chrome and Firefox and seem they both print ‘GMT’. I didn’t bother to test IE as I know the homepage doesn’t look pretty in IE and I am not a fan of IE 🙂

[Ray]

I was just about to comment that it’s probably due to unreliable connection of the battery, and saw that you’ve figured it out 🙂

[Ray]

The four pins on the RJ45 jack shouldn’t be the problem — they are for the two LEDs on the RJ45 jack, so not connecting them merely causes the LEDs to remain off and shouldn’t matter with the power circuitry. Since your VIN to GND resistance is still very low (29 ohm as you described), it suggests that there is still a shorting / damage somewhere on the VIN line. Try yo remove IC1 and see if that changes anything.

[Ray]

Hi Jerry, I suggest connecting each MOV between S0x and Gnd. This way, when lightning creates a high voltage across the solenoid wire and the circuit ground, the MOV will conduct to protect the circuitry.

[Ray]

The 1.1M ohm 5V-Gnd resistance is a problem. That most likely means one of five pins of the LM2596 switching regulator is not connected. You can take a look at those five pins under good lighting condition to see if any of them seems to be disconnected from the PCB pad. You can try to apply some solder to reinforce the connection if that’s the case. At any rate, this is an unusual case as all OSPi boards are fully tested. Again, if the problem cannot be solved, you can contact OpenLab for a replacement. Alternatively we can send you a replacement board directly, which will take 2-3 weeks to arrive.

[Ray]

The getweather.py script in GitHub is outdated. There has been some recent work contributed by Andrew that uses Yahoo Weather API. Check this thread for details:
viewtopic.php?f=28&t=184

[Ray]

Well, my original implementation uses this:
dstr=(new Date(lret*1000)).toUTCString().replace(” GMT”,””);
which prints out neither ‘GMT’ nor ‘UTC’. So I am actually a bit curious when you said it prints ‘UTC’. Perhaps the behavior is different on different browser. In any case, adding the time zone number is an improvement.

[Ray]

Thanks for your suggestion and the pointers. This has been discussed in other threads as well. If someone is willing to create an SD card image with all the necessary components installed, I am happy to find a place to host the image file. For the time being, I don’t have much time to work on this myself and hope someone can step up to have this done.

[Ray]

@dbrook: common off-the-shelf rain sensors are basically rain-activated switches. OpenSprinkler’s firmware handles rain sensor in the following way: if the ‘Use RainSensor’ option is turned on, and if the rain sensor is activated (this depends on the rain sensor type, either normally open or normally closed), then it disables all stations until the rain sensor is deactivated. In the situation you described, it will stop stations in the middle of a program, before the last station is completed.

Since the behavior of the rain sensor is implemented in software, you can modify the source code to implement any way you want.

[Ray]

Cool, thanks for the suggestion. I’ve integrated your change to the 2.0 and 1.8.3 home.js scripts stored on rayshobby.net. Will update the files on GitHub next. One minor comment: I assume you have to take out the ‘.replace(” GMT”,””)’ part, otherwise it won’t print either ‘GMT’ or ‘UTC’. Or perhaps this is browser-dependent?

[Ray]

Sorry to hear that. Here are my suggestions:
– Use a multimeter to measure the resistance between +5V and GND (there are multiple +5V pins, any of them is fine). It should be around 4.4 to 4.7 kilo-ohms. Also measure VCC to GND resistance, should be around 100 kilo-ohms. If not, it seems there is a damage somewhere on the board.
– If the resistances are correct, the problem is more likely with your 24VAC transformer, or the connection. You can use the multimeter to measure the *AC* voltage of the two wires of the transformer. Then make sure the wires are plugged into the screw terminal with good contact, and that you’ve plugged it into the ACIN port, not the COM port.

Let me know your findings. If the board is damaged or DOA, you can contact OpenLab for a replacement.

[Ray]

There is probably a shorting or some damage on the 3.3V line. Here are my suggestions:
– Measure the VCC to GND resistance. Should be at least a few kilo-ohms. If not, there is probably shorting somewhere.
– Start by removing ATmega328 from its socket. Use a flat screwdriver to carefully pry the IC up from the two edges, and be careful to avoid bending the IC pins.
– Measure the VCC to GND resistance again. If it’s at least a few kilo-ohms now, your ATmega328 is damaged. May be due to lighting, or if you’ve accidentally plugged the 24VAC power into the rain sensor port, that will fry the mcu.
– If the problem is not with ATmega328, proceed to remove other ICs, one after another, and measure VCC to GND resistance at every step, until you find which one is causing the problem.
– If all fail, the most likely cause is the MCP1700-33 linear regulator (3-pin, shaped like transistor). Try to clip it off the board, and replace it.
Let me know if these steps help.

[Ray]

Can you be more specific about the ‘4 of the pins near the edge of the board’? What component(s) do these four pins correspond to? Try to remove IC1 as well and see if that changes the VIN to GND resistance.

[Ray]

That can come out of a variety of reasons. For example, the programming is not connected properly to the mcu, the mcu is not up running (due to clock source problem etc). You can Google the error message ‘initialization failed, rc=-1’ to see potential causes.

[Ray]

To program a blank mega644 chip, use the following:
avrdude -c usbtiny -p m644 -B 250 -U lfuse:w:0xe2:m -U hfuse:w:0xd1:m -U efuse:w:0xfe:m
(in case avrdude complains the signature does not match, add a -F option to overwrite the signature checking)

Then you can flash a firmware with:
avrdude -c usbtiny -p m644 -U flash:w:firmware2.0.0.hex

[Author]

Posts