My solder station, a WESD51 had become increasingly flaky over the last year – at times, it would fail to turn on. Flip the switch up and down a few times, and the digits would light up and the iron would heat; flip a few more and it would remain inert. Sometimes it would work immediately, sometimes it would just not turn on at all. Every time I took it apart, it would trick me by eventually working again, only to fail when reassembled. To compress months of annoyance into one sentence: I seem to have fixed it by soldering some capacitors onto the oscillator crystal. Note added after the fact: Nope this didn’t fix it. See next post for grousing.
Generally, I like Weller irons. I had used the analog version of this model (i.e., no LED digits, just a knob to set temperature) for a couple years. During that time, I had sunk some funds into buying a variety of tips from teeny screwdriver up to the broad chisel that I use for soldering coax plugs. So, when it came time to move overseas, I found a 230V version online and ordered it. I had considered just putting the US model on a transformer, but thought it better to have a model directly grounded through its plug. Continue reading “Weller WESD51 Repair”
Propagation has been abysmal, so it’s time to hang out in garage and work on projects. One catch: the garage PC gave up the ghost about a month ago. The Windows 7 computer had been functioning for a few months as a wifi repeater that let me use other wireless devices in the garage. Unfortunately, it looks like a power spike may have taken out the motherboard. I have retired that PC, and came up with a replacement: a linksys wrt54gs router reflashed with DD-WRT firmware and hardware modifications to add a cantenna.
Last week I made a video about putting up the hex beam, and now that I have the video editing software, I made one about the wifi repeater bridge project. Making video is somewhat addictive, so I think there are more on the way. I have a ways to go in terms of production quality – maybe Christmas will bring a better video-capable camera.
On the subject of videos, my home club, the Vienna Wireless Society, is now posting videos of presentations made at the club.
There are a bewildering number of examples, libraries, and opinions about how to write routines to read values from a rotary encoder. I found one that that looked particularly elegant and easy to generalize, so adapted it to the Atmega328 that I’m using for my current project. I used a pretty popular rotary encoder, Adafruit #377, although I think the demo code would work with other quadrature encoders.
Most of the work in done in a relatively light interrupt service routine that receives information from both directional pins on the rotary encoder. I did not do any addition software or hardware debouncing and I didn’t notice any jitter or erratic behavior.
With wintertime (and remember, it is winter where I am) comes soldering – I can spend more than ten minutes in the garage without being drenched in sweat. I have a three week travel break before going on summer leave, so I’m trying to wrap up various projects. I have one project breadboarded and wired up to my arduino duemilanove, but I needed another arduino to work on another project in parallel, so I built one up on perf board: the Antananarivoduino.
The Antananrivoduino is about a simple as I could make it: an ATmega328, 16Mhz crystal oscillator, and 5V power. The one frill I added was a 28-pin ZIF socket that I’ve had kicking around for a couple years. The socket lets me seat a chip, program it, and eject it. It’s ideal for cranking out copies of a project or a burning a final chip that will go into an embedded project and won’t need any more programming.
This project arose from a sudden desire to stop inhaling solder smoke. I just got back from about three weeks of constant travel around Africa with a bad head cold, cough and sore throat. After being away, I was itching to solder up something, but didn’t want to push myself any closer to bronchitis. The solution: a fan to waft away the solder smoke.
I had a bunch of waffle fans removed from computers, but some of them proved very wimpy. One in particular was super-quiet and drew about 200 mA, but was barely able to move air. Since more is always better, I dug into the junque box and pulled out a bunch of fans that I have removed from a commercial server. They were small, but powerful. Alone, each drew about a half amp and when plugged in a 12V battery (or tapped into the bench’s 12V supply), would vibrate themselves around the table. If one fan is good… four is even better.
I created a wood frame with some channels made from angle aluminum. Since the metal impeller blades are like little razors, I stapled metal mesh over both sides. After assembling the whole thing, it occurred to me that all the fans should face the same direction. Luckily, the wood glue hadn’t set by that point, and I was able to disassemble part of the frame, flip around a couple fans, and screw the whole thing back together.
With four fans running full speed, this thing is loud and produces almost enough thrust to take off. So, the interesting part of this project is not the fan itself, but the fan controller. Any sort of rheostat or linear device that would ramp up or down the drive voltage would be very inefficient and produce a lot of waste heat, so I went with a variable pulse width modulator. Rather than reinvent the wheel, I used a design described on the PC Silencio website. Two sections of the quad op amp create a triangle wave, another serves as buffer for a voltage divider, and the final section serves as a comparator to slice off the triangle wave at the level set by the voltage divider. The threshold determines the duty cycle of the generated square wave, which drives the motor at full voltage.
I didn’t have a LM324 in DIP14 layout, so I substituted a TLC271 op amp, which worked fine. For the n-channel MOSFET driving the motors, I went with an STP16NF06 (my old standby, used in previous projects such as the TEAPOT and Texas Topper Amplifier). Regardless of setting, the mosfet remains cool since it’s essentially either on or off and not hanging out in the linear range at all.
In practice, running the fans at near the lowest setting works fine for pulling away smoke from routine soldering. Since the question will inevitably arise — I should mention that in order to solder together the fan, I used a breadboarded version of the PWM circuit to run the fans so I could solder them together — thus getting around this bootstrapping issue.
About the time I am wrapping up in the office, the sun has risen and people are getting into work on the East Coast of the United States. I leave early enough to get home so I can make teleconferences that start at 9 Eastern. I’d have quite a phone bill if I dialed direct, but fortunately, I can make most of these calls over the internet. Although Madagascar is a developing country, we have fiber to the house. The upstream connection isn’t entirely stable, but it is up most of the time, and throughput is surprisingly high. However, I still run into problems when the power blinks off.
The ISP has adequate backup power to maintain the connection, at least for some time, so this is my problem in the house. The house too has a backup generator, but there’s a couple seconds to shift over from mains to generator and vice versa. In that time, the houses’s router goes down and takes a minute or so to reboot every time power switches over. What’s needed is a UPS, but the smallest computer UPS here runs around $150. Also, it seems like a UPS is massive overkill when the problem is just to keep a router happy for about a minute.
I have a good stash of small lead acid batteries for portable ham radio use, and it occurred to me that I could probably use a couple in backup power supply projects. At this point, I’ve come up with two designs built of materials squirreled away in my garage. Schematics are below the cut for anyone interested. Continue reading “Internet backup power”
I envision having a few antennas: the hexbeam, the G5RV, lindenblads for 70cm and 2 satellite work, perhaps a vertical of some sort if I can figure out where to place it, and maybe some kind of beam antenna for six meters. To make all that work, I’ll need some sort of way to bring the lines in the shack and to switch among them. I brought four alpha-delta four position switches, which should be enough to both perform this function and switch the lines to the available rigs. With that intention, I laid the switches on the bench and drew out the wiring diagram.
However, I never got there; not yet, at least. After piling up some connectors and coax and wood, I realized that what I really needed in the garage in order to do this sort of work was some kind of background noise to keep me entertained. So, I pushed the very useful antenna switching project to the side and turned back to the computer that I had fried a few weeks ago by plugging it into the 230V while its power supply was set to 110V.
As a public service message, I feel obliged to share the following nugget of wisdom: before plugging a tower computer into a 220V outlet, reach around to the back of the computer and make sure the input voltage select is 220. Sounds simple, right? This isn’t something you have to think about for most laptops, which have dual voltage power supplies. They are happy plugged into either voltage and the power supply brick just works.
I’m not [arguably, perhaps] an idiot – I was aware of the switch. I just thought that the power outlet was off, but also I didn’t expect to encounter an issue until the computer itself was turned on. Wrong — ATX switching power supplies are always on. When I plugged the computer in 220V, there was a popping sound followed by smoke from the back of the unit. Never a good sign.
I pulled the power supply out, opened it up, and looked around. Nothing was obviously charred. My nose has lousy spatial resolution — it confirmed that something wasn’t right, but couldn’t help me localize the problem. I followed the wiring from the outlet inward. For a cheap supply, I was glad to see some decent capacitors on both live and return wires to ground, and across them. Also, some inductors to quell EMI. Next in line: the fuse. It had blended in because as a safety precaution, it was wrapped in heat-shrink. I cut away the heat shrink to reveal a white tube. I couldn’t see into the fuse, but my continuity tester showed it had blown. There was no fuse holder; the fuse was just soldered in by its leads, so I dutifully unsoldered it.
Back in October, I mentioned an open source keyer developed for the ATTINY 45 and requiring only a few components. The source code for the controller and an example schematic were uploaded to a repository. Subsequently, I decided to try my hand at producing a PC board. Rather than try some sort of printer-based method based on toner transfer, I wanted to try going a more professional route and having the PCB run off by a fab. Initially, I thought I’d have to go overseas and wait upwards of a month for my boards to come back, but I found a domestic fab that provides an amazing service for low volume prototype boards: OSH Park.
I had already laid the schematic out in Eagle CAD, which seems to be popular among hobbyists. Most of my components were available in off-the-shelf libraries, but I had to layout the piezo speaker as a custom part (although I started from another similar piezo speaker and just had to modify the dimensions). It took a while to get the hang of laying everything out on the PCB, laying down ground planes, routing the traces, and making the silk screen look nice, but after many hours with online tutorials, it all looked right. I ran some rules checks, and everything reasonable, as far as I could tell.
Next, I shopped around using the online check and quote tools available from a number of popular fab houses. To upload my design, in most cases, I had to send a zip file of the various Gerber layers — top copper, bottom copper, top silk, etc. But, for OSH Park, all I had to do was upload the Eagle board file itself. This makes a lot of sense, as the Gerbers are generated from that file, so all the information is already there in the board file.
The OSH Park site interprets the board file on the fly and provides a rendering of approximately what the board will look like when final. There are a number of options regarding cost, turn around, etc., but I opted to get three copies of my design for about ten dollars by agreeing to have my boards made as part of a larger run to take place in about two weeks from the date of submission. The way OSH Park makes prototype boards affordable is by merging multiple designs into one larger board and then cutting that board apart. Their cost is per square inch, and I had designed my board to be one by two inches — not bad considering that I used all through-hole components and did not go out of my way to pack them tight on the surface of the board. In fact, my design is a little generous in that I give a number of ground connection points, whereas only one is really necessary for the sake of wiring in external components.
From the time of submission until the envelope arrived, I was able to track progress of the order, so there was particular excitement on the day that I knew the package would be waiting in my mailbox. When I opened the padded shipping packet up, I found three little purple boards, just about identical to the rendering that was provided when I had uploaded the design.
The PC boards are excellent quality, with no alignment issues. The solder mask went where it was supposed to go, all the vias are functional, and the pads take solder well and have no tendency towards lifting. Components went onto the board without any fuss and when powered up, the board worked perfectly, the first time. Having verified that the design works, I’ve shared the board on the OSH Park website.
Now that this seems to be working, one option would be to run off more copies of the boards and do something with them — embedded keyers, stand alone kits, etc., but now that I’ve tried out designing a through hole board, I’m curious how much more compact the design would be with surface mount parts (and how much more difficult it would be to assemble).
As the days grow shorter with the approach of winter and activity shifts towards longer wavelengths, I took stock of my log and noticed that while I have racked up a reasonable number of contacts on 15, 20 and 40 meters, 80 meters lags far behind. I anticipate moving overseas in about six months, but before I go, I’d like to even up the score on 80 meters for this QTH.
My lack of contacts on 80m is a function of my antenna limitations — where I live, I can’t put a lot of metal in the sky. I have one outdoor antenna, a 43-foot vertical; the rest of my antennas are in my attic. My vertical antenna is, intentionally, not much to look at: a single, black wire that runs from the ground up into the top of a tree and is almost impossible to see from a few feet away. However, under the gravel of my backyard, there is a DX-Engineering radial plate. Eight radials spike out underground from that point under my property and into the adjoining forest. The antenna is fed by a coax line that runs underground from the house to that plate, where the center conductor feeds right into the antenna. The antenna was never very well tuned on any specific band, but it managed pretty well on 30 and 40 meters with either built-in or external tuners in the shack. With difficulty, it could tune 15 and 17 meters, and my LDG tuner could force it to work on 80 meters, but the amount of energy actually going out the antenna was pitifully small.
So, I decided that for this winter, the vertical would become a dedicated 80m antenna. The attic antennas can handle the other bands. My first thought was to make an inverted L for 80m, but the far end would extend off my property and would increase visibility of the antenna, particularly in the winter when there are fewer leaves for cover. I decided to work with the vertical radiating wire that was already in position, but to interpose a loading coil at the base.
Pete, K6BFA, lent me his MFJ antenna analyzer, and I measured the impedance of the antenna at the point where I anticipated the matching coil would be located. I measured at 3.7 Mhz, a bit higher in frequency than where I intended to operate and the complex component of impedance measured 278j. Since the antenna is a shortened radiator, this would be capacitive reactance, so -278j. I calculated the inductive reactance needed to null it out as xL = Xc/2*pi*freq, or 11.9 uH.
I had made a coil form from Schedule 40 PVC labeled “one and a half” inches, but measured its outer diameter as 1.9 inches. I wanted to wind a coil big enough for the about 12 uH needed above, plus extra so I would have some for shunt inductance (which I guessed would be around 2-5 uH). I figured 18 uH would be enough to have room to spare. Using the formula of n-turns = sqrt(inductance((18 * coil diameter)(40 * coil length)))/coil diameter, all values in inches, I came up with a three inch long coil with about 28 turns. This fit nicely inside the box that I had, so I went with it. Note that the coil shown in the box in the picture was my first attempt, and the coil turned out to be too small. There is a learning curve for this sort of thing, you know.
The coil was mounted on nylon screws and coils were made rigid with epoxy. The coil wire itself was some 18 gauge hook up wire that turned out to be too large for my protoboard, so I am glad it found a good home in the matching coil. The top of the coil goes to the antenna. The coax comes in the side of the box, and initially, I probed the coil to find a good matching point tuning at 3.7Mhz, intentionally above the CW portion of the band, where I wanted to operate. I found the optimal spot to bring the complex portion of the impedance to zero, and then played with the ground lead, trying to find a point lower on the coil that would yield lowest SWR at 3.560 Mhz, the QRP CW watering hole frequency. After playing with the placement of these two leads for a while, I was satisfied with the resulting SWR curve, which is shown below.
I could have shifted the curve higher in frequency, but I really don’t operate much voice, so I made the decision to optimize the antenna for CW and digital mode transmission at the lower frequency end of the band.
Back in the shack, I verified that I got the same measurements and switched the antenna through to my K3. The rig read the antenna as SWR near 1:1, so I made a couple test transmissions and worked stations in Hungary, Italy and Jamaica. I then turned power to 5W and worked a station in NY. It’s anecdotal, but the antenna seemed to be working fine. After calling CQ at 5W, I checked the reverse beacon network and noted that I was greater than 10 dB above noise as reported by stations in W1, W2, W3, W4, W5 and W7, which seems much better than previously.