[Ray]

Already replied to you.

[Ray]

Well, you can tell if the controller crashes when running a particular zone. OpenSprinkler currently does not have circuitry to automatically detect defective solenoids, so you need to find it out manually. I don’t have much experience with defective solenoid, but I guess there are quicker ways to test, such as testing the coil resistance (if the resistance is too low compared to other good solenoids, that indicates a problem).

[Ray]

The first thing to try is to unplug the common wire terminal (so the solenoids will not be activated) and see if the controller still crashes. If not, that means the problem is either a defective solenoid, or the PTC fuse under high temperature (as discussed in this thread: viewtopic.php?f=9&t=84). If the controller still crashes even after the common wire is disconnected, you should return the controller for a replacement. Send email to [email protected] for support.

[Ray]

I see, I thought you have 1.4.2u (which does not have 1N4148 but a 5.1V zener diode instead).

Anyways, it looks like you have the 1.4u. Is your PCB showing July 2012 as the date or Oct 2012? D4 only appears on July 2012 PCB. I grayed that part of the instructions out because I assumed anyone who has bought the July 2012 batch would have finished building the circuit by now.

[Ray]

When this happened, did you power the controller via USB or 24VAC? If you powered controller through USB, it doesn’t go through the power switch so the screen will be always on. If you powered through 24VAC, you should check the power switch and see if there is any solder joint shorting. The 24VAC goes through the power switch so it should cut off power when you switch it off.

[Ray]

I think this has been explained in the user manual. Specifically, in the online user manual, there is a section that describes ‘Understand how the programs are executed’:
http://rayshobby.net/?page_id=730#program
and the third bullet says ‘While a scheduled list is running, no other programs will be scheduled even if a match is found.’. Perhaps I should make it more explicitly by saying that ‘The other programs will be ignored’. I recall that the video also briefly describes the way programs are executed.

So if two different programs have exactly the same time settings, they will always be scheduled together and the stations will be serialized. If one happens to be a minute behind the other, it won’t get executed. The reason is mainly because if you push the second program behind the first one, it may get into conflict with a third program, and as a result all the programs will be pushed back. This can potentially lead to a stack overflow problem. I am not saying this will necessarily happen, but for robustness it’s important to make sure the algorithm does not crash under any circumstance.

More intuitively, if you have a number of tasks you have to do, and one is behind schedule, you could keep pushing tasks back but you might run into the risk of never get back on schedule, or the other solution is to simply ignore the task that’s behind. The current firmware uses the second approach, and so it is up to the user to make sure programs are not ignored unintentionally. The program preview feature is particularly designed to help you find these potential problems. If you find a problem in the preview, you can modify the program time settings accordingly.

[Ray]

Check if you have soldered RLED (100 ohm) resistor correctly.

[Ray]

Cool, that’s good to know. Thanks for trying this out.

[Ray]

Can you unplug the valves (either by unplugging the common terminal or the station terminals) and test again? It’s possible that the crashing may be caused by defective solenoid (also see the previous thread on this board). The other possibility is that if the ambient temperature is high, the resistance on the PTC fuse may increase significantly that blocks the power supply. If that’s the reason, I suggest replacing the PTC fuse or simply solder a wire across the two pins of the PTC to bypass it.

[Ray]

I agree with the concern, and I did think about this issue as I’ve heard one or two cases where users plugged the 24VAC into the COM terminal. I was considering using different connectors for different terminals, but in the end it’s difficult to source different parts. Your color coding suggestion is good, and I will post it in the instructions.

[Ray]

I think there are only Model A ($25) / B ($35), and rev. 1 / 2, there is no rev. B. Most RPis available on the market now are Model B rev. 2. I don’t know where you can get Model A in the US. I believe the difference between rev. 1 and 2 is that rev. 2 has two screw holes on board which rev. 1 doesn’t have.

[Ray]

@COsgood: if you have two valves open at the same time when sequential is checked, I assume it’s either because one is a master valve, or is defective and cannot close once opened. in the latter case, the defective valve (most likely shorting) will start drawing too much current that causes the voltage from the 24VAC transformer to significantly drop, hence the controller crashes.

As mrburns42 said, adding MOV would not help in this case because the problem is not caused by power surge.

[Ray]

Ok, I can’t see anything obviously wrong from the pictures. Probably the best solution is for you to send it back to me, so I can take a look at the board and send a replacement if it’s defective. Please email me at [email protected] to arrange for this. Thanks.

[Ray]

By the way, I didn’t find your attached pictures. Could you post them again?

[Ray]

Yes you can solder an MOV across the AC input pins and that should provide protection to transient voltages. Just make sure the exposed leads of the MOV don’t touch other parts (use an electric tape to cover the area underneath the MOV).

[Ray]

Do you have an oscilloscope? If so, you can connect it to any of the signal lines with ground (for example, CLK and GND) to monitor the signal.

If not, here is a hack that allows you to observe the signals: unplug the 8-pin cable from the OpenSprinkler Pi, and insert an LED (ideally with a current limiting resistor but I assume ok without) between a signal line with GND (for example, CLK and GND). Then you need to modify the code to insert delays (at least 1 second) between each toggling of the CLK line. This will allow you to watch the signals sent from RPi.

[Ray]

In the past OpenSprinkler has always come with a 750mA PTC resettable fuse. I’ve received a couple of report that in the summer when the temperature is extremely high, the resistance on the PTC fuse would increase to a point that it blocks the current supplied to the controller. So as you said, it poses a potential problem. Another issue is that these PTC fuses have a slow response time (typically a few seconds). So if there is a shorting on the solenoid, typically the transformer voltage would drop instantly before the fuse even catches on. So in practice I don’t know how useful it is. In the most recent kit I’ve removed the PTC fuse. Probably a better solution is to use a fast-action fuse. Unfortunately there is not enough space on the board to fit such a big fuse.

[Ray]

Every board has been tested with a Pi before shipping, so the only thing I could think of is perhaps the shift register is not getting signal from the RPi correctly. Not sure what caused this, but that’s the most likely cause. The RPi sends signal to shift register through 4 pins: NOE (output enable, active low, usually pulled high, but is set to low when a program is running), LAT (latch), CLK (clock), and DAT (data). In particular the NOE pin should be set low when a program is running, otherwise the output will be disabled and hence no voltage.

Could you post a high-resolution picture of your setup?

[Ray]

With Pi Rev B, you need to change one pin number in the Python scripts. This is explain in the README.txt of the demo folder. For example:
https://github.com/rayshobby/opensprinkler/blob/master/OpenSprinkler%20Pi/software/demos/manual_buttons/README.txt

But even without changing this, the self-test program should still work. Were you not able to get any output voltage even running the ospi_selftest?

[Ray]

Sorry, I mis-read your question: you mean you can’t get a connection even by using the IP address? What is your OpenSprinkler’s IP address? Click the first pushbutton (B1) should show the IP address, and second pushbutton (B2) shows the router’s IP address. Also, on the LCD do you see a question mark at the end of the second line?

[Ray]

This means the controller did not successfully register a host name. This is not uncommon and there is nothing to worry about. I’ve tested several different routers and WiFi adapters, for some it works, for the others I have to use the IP address. Since the software uses JeeLab’s EtherCard library, you can probably post the question on JeeLab’s forum and see if there is a solution.

[Ray]

OSPi does not have a microcontroller, so there is nothing to re-program. Maybe you mean to re-program the Raspberry Pi?

For part list, please see the comment here: http://rayshobby.net/?p=5339#comment-1089

[Ray]

@Donald: 2.0 development is under way. I have the circuit design ready, and as you may have noticed from this blog post:
http://rayshobby.net/?p=4914
the plan is to release it with the new custom molded enclosure designed by SeeedStudios. The mold design is close to be finalized, but there are still tweaks they need to make. Apparently designing an enclosure that meets industry standard is not an easy thing: there are many engineering and mechanical considerations. I don’t have a reliable estimate of when this will be ready. I would say in a month time?

But during the time to transition to the new enclosure, I will start offering the development version of 2.0 that goes with the current enclosure. It will be similar to the picture you see on the blog post.

[Ray]

I am considering preparing an SD card image, which has all the necessary packages installed and the programs ready to go. Once this is ready, hopefully a user just need to get an SD card, copy/flash the image onto the SD card, insert it to Pi and starting using the programs.

[Ray]

Mark, just noticed your last question: you need to download it to your Raspi. You can use
wget xxxx
where xxxx is the download link.

[Author]

Posts