The title pretty much sums it up. My first prototype build for the HydroGadget was a success. I have a decent list of issues that need to be addressed for the next build, but for torture testing, this should work out fine. I think the plan is to have this prototype build on display at the SynShop booth at First Friday on May 3rd, and at the Las Vegas Science and Technology Festival on May 4th. I should be manning the booth on the 3rd, but my partners in crime will be manning the both on the 4th as I’ll be in class.
Download: HydroGadget Rev 0.F Bug List
Started assembly of the HydroGadget Rev 0.F. The parts that I was most worried about fitting correctly on the PCB are fitting perfectly. Sweet! I’ve been burnt in the not so distant past by not paying close attention to mechanical tolerances on datasheets. However, I did manage to ignore the rather simple and mundane. When doing the layout I laid out for 1/8W resistors, but put 1/4W resistors in my bill of materials. Since I already have the parts, I guess I’ll jam these big boys in and make changes in the next revision of the BOM. Other notes so far: The fuse holders have a much higher profile than what I was thinking. I’m trying to have the terminal strip be the tallest item on the board to make the design for the enclosure easier, so gotta find shorter fuse posts. Mouser had a ton of them, I just need to sort through them. Also need to move a trace that is right next to the terminal strip about 25mils or so north. I didn’t realize the bottom of the terminal strip I selected had exposed conductors going all along the bottom. Typical first PCB spin problems…
Three beautiful purple prototype PCBs of HydroGadget Rev 0.F came in the mail today. Simply cannot beat the value OSH Park offers. I do need to track down why DipTrace didn’t tent my vias on the gerber export. The preview within DipTrace shows the vias tented, but the gerber files it exports has the solder mask removed from the pad like it was a normal through hole. Not that big of an issue in this circumstance, but I would like to track this down for future projects where it might very well matter. I added these photos to the rev 0.F online document package. Now on to finding what I goofed on this PCB spin…
I really meant to get this finished and published a month ago, but life got in the way. Put in a whole lot of extra hours producing a special one time Cirque du Soleil production in our showroom to benefit “One Drop“, a world water access charity. The event was a huge success bringing awareness and financial support for One Drop, but it was a lot of work to get it there.
I tweaked a couple of things on the design schematic to make testing the performance of the circuits a bit easier. I have a handful of header and shunt sets along the power paths to allow me to interrupt the trace and measure current passing from one component to the next. I’m becoming less and less fond of my power supply design, so don’t be surprised if you see something completely different on the next revision. I also split up the main filtering capacitor into three separate smaller capacitors. Something smells fishy about my ripple calculations and this allow me to easily add filter capacitance in stages and see how my circuit responds. If I do end up using this basic power supply design I want to know how close I am to the margins. There is also an additional LED that is going to indicate the health of the programs running on the Raspberry Pi. If everything is healthy it should blink briefly every 5 seconds or so. If the LED is stuck on or doesn’t blink, we know something has crashed on the Pi and an investigation is in order.
One issue that we are trying to tackle is how do we initially configure the HydroGadget network settings and how would adjust those settings if we changed the name or password of the wireless network it attaches to. We decided that a network settings reset switch would be needed to put the HydroGadget into an ad hoc wireless mode to allow a computer or phone to connect directly to the HydroGadget. Once the HydroGadget is told what network to attach to and what password to use it would then go back to a client mode and connect to the wireless network like any other device would. One of these classic chicken and egg problems. A simple tactile push button and some resistors and we now have a magic reset switch. At the request of one of our programmers, I added a simple serial in/out port to allow for debugging code.
The layout is definitely a work in progress. I’m pretty sure the design of the output side is fairly solid, so I spent a fair bit of time getting that section of the layout nice and clean. As I said earlier, the power supply concept might very well change completely, so that side of the layout isn’t nearly as polished. I had to back mount the filter capacitors to try to keep the profile reasonably tight so designing a case for this wouldn’t be a complete disaster. Since I had the space for the LM317 to mount to the board, I poured a large ground plan on both sides to act as a little heatsink and riddled the section directly under the LM317 with vias to help transfer heat to the back plane.
I fully expect there to be at least one big mistake on this board revision that will require all types of fun hacking. I’ll be sure to post pictures of my butchering. From talking with seasoned professionals, you never get it right on the first spin. I sent in my gerber set tonight to my friends at OSHPark so I should have three boards show up in my mailbox in about 14 days. I have included all my source files in the link below. Stay tuned for the next update.
Here are some shots hot off the CCD from Maker Faire Las Vegas debuting the HydroGadget for the first time to the public. Turn-out to the faire was quite nice considering it’s a first for Las Vegas. We had a lot on interest in our project from faire visitors as well as other makers. Give us a week or two to recover and make up for lost sleep and we will be back to work on HydroGadget.
The main blog for the overall HydroGadget project is located at HydroGadget.org
There are a handful of websites associated with the HydroGadget project. The official blog that covers the overall project as a whole is at hydrogadget.org. The software side of the project is housed at our software repository on GitHub at github.com/hydrogadget. You can find all hardware aspects here at SmokingCircuits.me. We will strive to make sure important updates hit all three of our sites.
It’s alive! Sometimes protoboard builds are not exactly great looking. This power supply board is a good example. I’m rather pleased however, how the output board came out. I couldn’t locate a fuse of the right size so I just dropped a machine screw I had laying around so I could run tests. The power supply board passed a 1.5A load stress test for a full hour with no signs of distress. Should be ready to go for Las Vegas Mini Maker Faire!
Power supply board:
Below is a link to the HydroGadget Rev D schematic. There is a PDF version for easy viewing as well as the native source file in DipTrace. This is still very far from a finished project. The goal of this revision was to prove a working concept and have a working demo for the Las Vegas Mini Maker Faire. Over the next few months I’m going to work on the design to improve circuit simplicity and reduce the component count. My goal is to have the next public release ready sometime around the end of the first quarter or the start of the second. Feel free to comment on my design, especially if you have some constructive criticism. I’m a hobbyist that just recently started my formal electronics education, so there are all types of gaps in my knowledge.
I was approached by a friend at a SynShop meeting about some problems he was having power supply for a Raspberry Pi project he was working on. One thing lead to another and now I’m running the hardware side of this project so the rest of the team can focus on the software side. The concept of HydroGadget is an internet connected home irrigation timer and controller. What separates HydroGadget from other products on the market is the ability of the unit to connect to outside servers and change its operation accordingly.
For example, the HydoGadget could connect to a local weather database and see that last night it rained a quarter inch and that it doesn’t need to water this afternoon. Local watering restrictions sometime dictate what days and times you are allowed to irrigate based on the time of year and your local address. Here in Las Vegas, these allowed days change every few months and it’s easy to forget to change the irrigation timer. Watering at the wrong time can result in water district fine, and it’s just rude given our limited water supply for our region. Put your address into HydroGadget once and it automatically changes your watering schedule as needed to be in compliance.
The HydroGadget is going to be a add-on board for the Raspberry Pi, a extremely low cost and low power Linux based micro-computer. I don’t even pretend to know enough about programming the Raspberry Pi; that’s the specialty of other team members. The Pi will have a little USB based WiFi adapter that would connect to your home WiFi network and allows interaction with outside servers as well as the web based control interface you use to configure it.
Early on we decided that we wanted this project to be simple enough that a person relatively new to soldering could successfully assemble the board with pretty basic tools and supplies. This ruled out using any surface mount component and using through hole exclusively. We also wanted to limit the number of manufacturer specific or single source components we used to make it as easy as possible for people to find parts and avoid the hassle of discontinued components. All but one of the components are “jellybeans”; meaning they are a generic part that can be sources from many different vendors and manufacturers.
The hardware concept is actually quite simple. Power for the HydroGadget is taken from a 24VacRMS transformer that are standard for irrigation timers as the water valves are typically 24VacRMS. The 24VacRMS passes through four diodes and a bypass capacitor produce a rectified signal. This feeds a LM317 variable regulator configured to limit the output to about +34Vdc. A tiny high efficiency dc to dc switching power regulator then produces our +5Vdc system power. I found that even though the transformers typically used all specify 24VacRMS, I found that with only a small load, the secondary voltage can be substantially beyond 24VacRMS. Little RMS to peak math and found that a typical transformer can spit out as high as 41VacPeak. The little dc to dc converter I found is great, but can only handle 36Vdc input. Enter the ever flexible LM317. With two resistors I can set the output voltage to 34Vdc, which the dc to dc converter can easily tolerate. The LM317 has a useful little quirk that it is stable even when the input to output voltage difference is below the dropout voltage of about 2Vdc. When the input voltage is below 36Vdc the output voltage is equal to the input voltage minus the dropout voltage. So in essence, the LM317 is acting like a voltage limiter for the dc to dc converter.
The output board is not nearly as interesting. A ribbon cable connecting the output board to the Raspberry Pi feeds the +5Vdc system power as well as the +3.3Vdc logic out signals from the Pi. The logic out signals are run through a eight channel Darlington pair IC to minimize the load on the Raspberry Pis GPIO pins. The Darlington pair IC drives four optocouplers with triac outputs. These weak triacs then drive a much higher capacity triac which actually switches the 24VacRMS on and off to the valves. Besides the inherent safety advantage the optocouplers provide, it also adds flexibility to our design to switch different or higher voltages if someone wants to do that modification. I’m very curious to see what people end up hacking HydroGadget to do.