tools of the trade

In the past few months I have been doing quite a bit of reading in my spare time in Jerry Sevick’s (W2FMI) book Understanding, Building, and Using Baluns and Ununs. This text is readily available from a few sources including Amidon Corporation, from where I picked it up (as well as a few toroids). My main reason in delving into Sevick’s tome was to increase my knowledge and eventually save a little money by constructing my own, high quality baluns, ununs, and RF chokes at home.

My current antenna setup consists of a ground-wave vertical mounted on top of a split-rail fence behind my townhome.

Lack of space, trees, and presence of a HOA have kept me from erecting a large loop antenna, as I would like, or exploring other more conspicuous options. One issue with the use of the vertical has been the “RF in the shack” problem–especially on digital modes. In some instances, I would transmit a PSK31 signal on 20m and the SignaLink USB as well as the software controlling it on my computer would freeze up and require a hard restart in order to remedy the situation. I hadn’t been bitten by any RF on my microphone yet, but this was enough reason for me to explore a solution.

preparing the bifilar wires

I thought it was time to add an RF choke to my antenna system. After reading Sevick, receiving some helpful advice on the eHam.net Elmers forum, and doing some other reading on the internet, I decided to go with an FT240-43 toroid which would have adequate (i.e., greater than 5000 ohms) choking capability throughout the 10m through 40m bands, where I would usually operate. Others had suggested a -31 mix core, which would offer greater choking at lower frequencies; however, I don’t expect to operate much on 60m, 80m, and 160m at this time. The -31 mix core doesn’t provide as much choking over the 10m through 40m bands as the -43 core, so that made my choice a little easier.

In addition to the choice of the core, I also had to decide whether I would wind the toroid with coaxial cable or enameled copper wire. I made the choice to use #14 enameled copper wire because it seemed easier to work with. Using coax usually requires the use of Ty-Rap cable ties to hold the coax close to the core; the #14 wire would stay where I put it on the core.

One thing that seems left out of the description in Sevick’s text is the length of wire needed to cover a certain number of turns on a particular size core. Thankfully, Diz W8DIZ provides a link to a very helpful site with this information–toroids.info. This calculator found that I would require 26″ of wire to create 10 turns on the FT240-43 core. I didn’t want to run out of wire, so I used 33″ (don’t ask me how I chose that number) of wire instead. It turned out that 26″ (well, maybe 28″) would have been sufficient.

wound core

Sevick states that the impedance of bifilar #14 wire is around 45 ohms. In order to increase the impedance to 50 ohms, I would have to wrap one of the wires in Scotch #92 tape. After wrapping the wire in #92 Scotch I placed both wires in a Panavise to hold them together as I bound them together with Scotch #27 tape. This is an important step because it keeps the two wires from traveling away from each other as they are wound around the core.

I prepared the core by covering it in #27 Scotch tape. This helps the bifilar wire stay put on the core and also protects the wire from any sharp edges that may be present on the core, which could cause nicks or shorts.

The RF choke can be wound a few different ways. You can wind all of the turns on 1/2 of the toroid (closely spaced), wind the turns spaced evenly around the toroid, or wind the turns evenly around the toroid but utilize the “crossover technique” developed by Reisert W1JR. Winding the turns evenly spaced around the toroid leaves you with the input and output at the same point (nearly) on the core. Physically, this is not desirable. Using the crossover technique or winding the turns over 1/2 of the core allows for the input and output to be on opposite ends of the core. Sevick states that the electrical properties are similar if you choose either of these techniques. I chose the crossover technique since it seemed easier and more “economical” (acc. to Sevick).

preparing for soldering

Once the core was wound, I had to find a suitable enclosure. I decided to use a Carlon 4″x4″x2″ junction box. These are readily available at places like Lowe’s for less than $7 each. They have a nice rubber grommet at the top of the box which helps keep out moisture. I also found them very easy to drill with my Unibit stepped drill bit. I used SO-239 connectors which require a 5/8″ hole if mounted on the inside of the box, as I chose to do. After placing the SO-239 in the 5/8″ hole, I used my Dremel to drill holes for the 4-40 hardware to hold the connector in place. After placing one bolt in the first hole, I drilled the second hole. I mounted the connectors diagonally rather than square (this makes more sense if you look at the photos).

finished product

The next step was to solder the wires from the wound toroid to the SO-239 connectors and close up the box. After a few snips here and there I had a snug fit. The soldering iron did its job and I was done with my work.

I took the finished product to Steve KZ1X’s for final check-out. He used his AADE L/C IIB meter to measure the inductance at 136 uH. Using a simple formula, the calculated impedance across the intended frequencies of coverage was much greater than the 5000 Ohm goal I had at the beginning of the project. I hope to analyze it further and provide more detailed (i.e., measured instead of calculated) impedance values across the bands of interest in the future.

More photos of the project may be found in this album on Picasa.

I may warm up the soldering iron this evening and fire up the oscilloscope which W3AHL has graciously allowed me to borrow in order to check the voltage from the frequency synthesizer. According to the specs, the SA612AN expects a 200-300 mV signal from the oscillator. The stock output is much higher than that. W3AHL had suggested that I use a Pi attenuator to reduce the voltage from the synthesizer. I followed his initial advice, but without an oscilloscope I wasn’t able to check the voltage. Before I had the chance to test it, I had received this message from him:

The only problem is the 10db attenuator is wrong.  Last night I realized we were measuring voltage — I calculated the attenuation needed based on power ratio, which is what I normally use for RF circuits.   So just to make sure I got it right this time I verified what was needed using the 8924C’s signal generator and a 30 dB attenuator I had.

So to take the 2.2 volt P-P signal (unterminated — it would be 1.1 volts if the PSK-20 board had a 50 ohm terminator resistor on its end of the coax) down to the desired 250mv. signal, requires an 18.9 dB attenuator.  This would require two 62.8 ohm resistors to ground and a 217 ohm resistor in series between them.   Or as close as you can get in standard values.

The current 10 dB attenuator should give 700 millivolts output — a tad too much.

Again, I’m glad to have the experience and expertise of folks like Steve W3AHL to help with projects like this one. I’ll post results when I have a chance to test them–hopefully this evening.

Yesterday I ordered a low pass filter kit from Dieter “Diz” W8DIZ, who runs an awesome site/store called KitsAndParts.com. I explained to him what I was trying to do, and he offered to help me create a low pass filter with a cut-off frequency just above the output of the frequency synthesizer (around 5.3 MHz) to help transform the square wave from the synthesizer to a sine wave (which is preferred by the SA612AN).

John,
I made you a custom LPF for 5.3 MHz
Need to calculate the turns on the T50-2 cores
Caps are 470p and 1000p
All packed and ready to ship tomorrow AM
-Diz

Diz has always been helpful and mindful of his customers’ needs. This can be clearly seen in his reviews on eHam. Hopefully the parts will be in by the weekend and the next phase of the project can be tackled before the beginning of the work week.

Steve W3AHL, a member of our local club, mentioned to me a few weeks ago at our regular monthly meeting that he had a ton of lab-grade test equipment at his house. Immediately, a light went off in my head, and I made a mental note to contact him whenever I was preparing to do the final checkouts on the PSK-20 project.

Yesterday, I headed down South to Chatham County to visit him (about a 15 minute drive from my home in Carrboro). We spent the better part of 3 hours with his spectrum analyzer and oscilloscope trying to figure out the answers to the following questions:

1. What is the exact frequency of my Si570 oscillator?
2. How far can I pull the 9 MHz LO down?
3. What additional adjustments are needed in order to use the Si570-based signal generator as the second LO?

The answer to the first question was fairly easy to determine. After applying power to the Si570 frequency generator kit and allowing it to sit for 20 minutes or so, the output frequency was 9,999,992 Hz. Its nominal frequency is 10 MHz, so that’s not bad. Saving this value to memory space “000″ in the menu allows the kit to generate more accurate frequencies than if you work with the assumption that the nominal frequency is equal to the actual frequency.

The answer to the second question is important because that will determine if the crystal I have will allow me to operate on USB rather than the LSB (which the kit is designed to operate on). This odd configuration may be overcome in some software packages. At the end of the day, I’d like to be able to operate on USB (the traditional sideband on 20m); however, if this ends up being a costly choice (dollars-and-cents-wise) I may bypass it. Measuring the frequency of the 9 MHz LO while adjusting the trimmer capacitor C38 helped me determine that the lowest frequency of the oscillator is 9.021 MHz. This is definitely not as low as I’d prefer for the generation of the USB signal. Inrad sells USB crystals for the frequency of 8998.5 kHz, but their cost is $12 plus whatever shipping and handling may be involved. Not exactly the least expensive solution. So, the solution for now is to accept LSB as my sideband of operation on 20m digital modes until such point that I feel like spending what will amount to at least $20 to obtain this new crystal.

The final question to be answered of the day was how to incorporate the Si570 frequency synthesizer into the kit as the second LO. Previously, I had thought about using high-side injection at 23 MHz (instead of 5 MHz) “just for fun” and because other folks on QRP-L had suggested it might be a cleaner signal. After speaking with Steve KZ1X I decided that it might be best to stick with the low-side injection method since using high-side injection may require additional changes to the circuit which I may not want to tackle at this point in my learning.

Going with this thought, Steve W3AHL fired up his oscilloscope and began measuring the voltage output of the synthesizer and comparing it to the voltage required by the SA612AN Gilbert cell mixer. It turns out that the peak-to-peak voltage is much higher than what is needed/tolerated by the SA612AN. Steve W3AHL helped design a 10:1 Pi attenuator using commonly found resistors from Radio Shack. In a perfect world (at least, according to the calculator he used), I would have a pair of 96 ohm resistors and a single 71 ohm resistor in the circuit; however, Radio Shack’s selection of resistors is limited to 100 ohm (for the 96 ohm) and 68 ohm (for the 71 ohm). Not a bad compromise at all!

The neat thing about this attenuator circuit is that it will fit on the PC board very easily. Recall that I did not install any of the 5.068 MHz oscillator parts. This has left open C43, C44, R47, R48, and R49 (as well as a few other pads). The Clock + (in the circuit) and Clock – (to ground) from the Si570 kit were fed using RG-174 coax at C44 and a wire jumper was added at C43 to make the connection between C44 and R47. The Pi attenuator was created by adding one 100 ohm resistor at R47 and at R49 and one 68 ohm resistor at R48. I haven’t had a chance to test this yet, but it should provide the correct amount of voltage for the SA612AN (200-300 mV). It still may be the case that I need to add a LP filter to create a sine wave from the square wave the Si570 creates, but I’m going to see what happens when I power it all up before I go that route.

Our regular monthly club meeting is this evening. I’m planning on bringing the kit to the meeting for a little intermediate show-and-tell and eventually giving a full presentation of my project to the club in February or March.

As of today, I have completed all but the last two steps in the assembly of the kit. The manual is broken down into “group builds” (e.g., Group 1 assembly installs the DC power circuitry, Group 2 assembly installs the Transmit/Receive switching circuitry, etc.) which help the kit builder have a sense of the purpose of each set of capacitors, resistors, transistors, and the like. Unlike Elecraft kits, which usually have an alignment or “check out” after each step of assembly, there are no such resistance or voltage checks for this kit. So, you must go on faith alone that there are no cold solder joints, misplaced parts, or nonfunctional devices until after assembly is complete. The first “smoke test” is truly that.

In the Group 3 build (the 9.000MHz and 5.068MHz oscillators and related components) I left out all components connected to pin 6 of U7 (SA612AN) since I plan on connecting the signal generator there in lieu of the 5.068MHz crystal. It may be the case that I will need to add some of those components once I determine their necessity, but there is plenty of physical real estate on the board for these parts to be installed at a later time. In face, Dave Benson states that “real men” can install in whichever order they wish, ignoring the “group build” guide laid out in the manual.

The last two portions of the construction include stuffing and soldering the transmitter bandpass filter and driver stages as well as the transmitter final amplifier to the PC board. I hope to finish those either tonight or tomorrow and then take the final product to fellow OCRA member Steve W3AHL’s home to tune the sideband generating circuitry and obtain an accurate measure of my particular Si570 part two days from now. After that, I should be ready to place the kit in an enclosure and start playing around with it on the air!

Once I complete the build, I’ll upload some photos of the board after each “group build” for viewing.

Despite my best efforts to botch the assembly and initialization of the device, I successfully completed it over the weekend. Partly due to me breaking one of the rules of kit-building (i.e., no building kits when you are sleepy or tired), I soldered the V+ and ground from the power supply to the output locations. During check out, the unit wouldn’t power up. A quick email to Kees K5BCQ asked me to check the voltage on the regulator–which led to my realization that the V+ and ground weren’t connected to the regulator. <sigh> Lucky for me, the location of a few capacitors kept me from frying any ICs. Once I connected the power to the correct points, and performed the initialization steps (holding down the push-button while applying power), everything is working as expected.

The Si570 comes programmed with its initial frequency from Silicon Labs. I requested 10 MHz initial frequencies. The initial frequency of the Si570 is set in the menu of the kit in menu item 000. While 10,000,000 Hz should be fairly accurate, I’d like to measure the exact output at the indicated frequency to make sure that it is not off by a few hertz in either direction. The accuracy of the start-up frequency of the Si570 in the menu item affects the frequencies generated using the controller, so this is key to its operation.

The kit is flexible in the voltage required for its operation. I’m thinking about adding a 5V regulator from the main power jack and feeding the display/generator rather than having to drill another hole for a separate power supply for the device. Using a wire jumper (rather than a 1/4W resistor) near the voltage regulator, allows for the use of power less than 11V.

In addition to the printed instructions provided with the kit, I followed the instructions on Clifton Laboratories’ site which were very helpful. Thanks for providing this excellent resource, Jack!

Next–on to building the PSK-20!

In 2010 I will be traveling a bit for my job and may find myself in lonely hotel rooms at night. Rather than spend the evenings flipping through channels on the television, I would prefer to get on the air and make some contacts. Given the fact that I travel with a laptop computer and often travel light, I thought it would make the most sense to try and bring along a transceiver which would allow me to get on the air and operate some of the digital modes. Dave Benson’s PSK-xx kits fit the bill.

I chose the PSK-20 because of the high activity on 20m PSK31 and because the other kits are for bands which would require an antenna of greater size than I would prefer to manage on business travel (PSK-30 on 30m, PSK-40 on 40m, The Warbler on 80m).

I’ve previously operated PSK31 on my Yaesu FT-817D with a Tigertronics SignaLink interface and been quite successful, earning the QRP-ARCI 1000 Miles Per Watt award (see post here about those adventures). Using a transceiver with a built-in interface would save space and weight when they are at a premium.

I ordered a PSK-20 rig in the summer of 2009 and finally received it in the fall, but have yet to begin the project. With increased confidence in my homebrewing abilities and greater understanding of theory, I have set out to modify the PSK-20 kit to tune the entire 20m band and allow me to reach the calling frequencies of digital sound card modes other than PSK31.

The PSK-20 rig itself is an interesting rig. It has a ~4KHz bandwidth centered around the PSK31 frequency for 20m (14.070MHz) and operates on the non-traditional lower sideband (LSB). The sideband is of no consequence for some variants of PSK31 (e.g., BPSK) but matters more for others (e.g., QPSK).

According to Dave’s documentation about the receiver…

U1 is fed with a 5.07 Mhz Local Oscillator (LO) signal and converts the incoming 14.07 Mhz received signal to the 9.00 MHz Intermediate Frequency (IF).

and

Product detector U2 receives the 9 Mhz IF filter output and multiplies (mixes) it with a ~ 9 Mhz LO signal.

Also (for the transmitter side of things)…

Single-sideband filtering is performed by a second filter comprising Y7-Y10 and related capacitors. After passing through the transmit IF filter comprising Y7-Y10 and related components, the signal has been reduced to a 9 MHz SSB signal. Its output drives 2nd mixer U7. This mixer is also driven by a signal from the 5.07 MHz LO, Colpitts oscillator Q9 and associated components.

The output of this mixer consists primarily of both the sum (14.07 MHz) and difference (3.93 Mhz) frequencies applied to the mixer. Q7 is an emitter follower used to buffer the high-impedance output of the mixer. U8 is a Monolithic Microwave IC (MMIC) and provides approximately 12 dB of gain.

My plan, as it currently stands:

1. Pull the 9.0MHz local oscillator LSB signal to USB by adding a level of inductance in line with the trimmer capacitor already on the board
2. Replace the 5.068MHz crystal and its supporting circuit with a frequency-agile signal generator
3. Mount the entire rig, including a digital frequency readout, in a Ten-Tec enclosure (the enclosure that Small Wonder Labs provides for the PSK-xx rigs is not large enough to accommodate the size of the digital frequency readout and dial)

In order to make this a fully portable digital station, I’ll need to bring along an antenna (I’m thinking either a 1/2 wave wire antenna cut for 20m) and perhaps an antenna tuner. The tuner, should I need it, will be filled by my Elecraft T1. It offers small size and excellent performance that will be appreciated when packing my bags for my business trips.

Should be an exciting few weeks. My first trip will occur in the third week of January, and I hope to take the rig along with me. First, to construct the signal generator of choice–the Si570 Controller and Frequency Generator Kit #2 by K5BCQ and K5JHF.