The Elecraft K3 is a Joy on RTTY

K3 - Right in the operating position

I built the K3 just after Christmas. Since then, it has been a joy to operate in many ways. Having just completed my third RTTY contest this year, I'm especially pleased with how it operates on RTTY.

Although I operated packet radio back in the 1980s, I never got a make a RTTY contact until 2005 - using my Elecraft K2. I built an audio interface using a couple of audio transformers and some resistors.

In many ways, the K2 is a great radio. Gary Breed K9AY told me that the K2 was, "a simple radio, superbly executed." While I do enjoy the K2, it is not at it's best running RTTY or any other digital mode. It suffers several limitations:

• Duty Cycle - The K2 is designed for a 50% or less duty cycle - CW or SSB. Without supplemental cooling, you can't run the K2/100 much more than about 25-30 watts. I added a small fan on top of my K2, and could safely run 50-75 watts of RTTY without overheating. The K3, of course, is a 100% duty cycle rig - no problem running 100 watts for hours. The case doesn't even get warm.
• Tuning Rate - Because of it's design, the K2 tunes in roughly 9 Hz steps. While the display shows the nearest 10 Hz at all times, the firmware selects the closest tuning point. As you turn the dial, it's not entirely smooth, sometimes jumping twice as far. With the finicky tuning of RTTY signals, you'll notice this.
• Frequency Readout - I put this one in my K2 Wish List. The K2 frequency doesn't account for the mark frequency, and thus reads 1-2 kHz high, depending on what audio tones you are using. While some contest software can compensate for this, the K3 works correctly by showing the mark frequency on the display.
• Variable Transmitter Gain per Band - this is the most annoying thing about using the K2 on RTTY. Because of the way the K2 transmitter gain control system works, you have to adjust your audio levels on each band -- very high levels on the higher frequencies, and extremely low levels on the lower bands. Every time you change bands, you must adjust the transmit audio level. Every time. The K3, on the other hand, calibrates itself so that it has the same gain on every band. 

But the best part isn't that the K3 can run flat out 100 watts, tune smoothly, display the right frequency, or make it so you don't have to mess with the transmit audio level in the middle of the contest. These other features of the K3 really enhance RTTY operation:

• Variable-Bandwidth IF - My K2 is programmed with a variety of bandwidths for RTTY: 1000 Hz, 500 Hz, and 300 Hz. But this isn't nearly as convenient as being able to select any bandwidth from 100-4000 Hz. On a quiet band, a bandwidth from 600-800 Hz works well. When the QRM gets tough, pulling down to 300 Hz puts it in its place. The K3 allows you to pick exactly the right bandwidth for the situation.
• Dual Passband - RTTY contests can get downright ugly. Sometimes strong stations are interlaced on top of one another, so the mark or space frequency of one station is right between the tones of another station. This situation does no good for anyone, as it makes it nearly impossible to copy either station. I'm convinced that this is largely due to the over-use of  dual passband filtering. This type of filter puts a huge notch between the mark and space tones. It's not useful for S & P work, because if you are slightly off frequency, you won't hear stations. And, it's not terribly useful for calling CQ, either, since someone can easily move in between your tones and you won't hear them. However, every once in a while, this filter makes the difference between completing a QSO and losing one to the QRM. Just don't use it all the time, OK?
• Fine Tuning - While the K3 tunes in perfect 10 Hz steps, you can select fine tuning mode with 1 Hz steps. I used this all the time as default for CW and RTTY and move it. It makes tuning across the band take longer, but it's easy enough to switch to 10 Hz steps when you need to.

So, every time I changed bands, tweaked out QRM with the filter or passband tuning, marveling at how cool the rig is running pumping out 100 watts hour after hour or just simply tuned stations in -- I'm consumed with joy over this fine radio. Definitely worth it.

K3!

K3/100 build and under the Christmas tree.

I have the best wife in the entire world.

When it came time to make our Christmas lists this year, I had mentioned that it had been ten years since I bought my last ham transceiver -- I finished the KPA100 for the K2 in the fall of 2002. And while it might be time to look at something new, modern transceivers are not cheap. In the end, I decided to ask for contributions toward the "K3" fund -- money I would use to buy an Elecraft K3 at some point in the future.

Before Christmas, my wife had hinted she had gotten me something expensive. She would tease me that she had purchased a Cessna 140 or a Harley-Davidson. She did it often enough I wasn't sure what to believe.

Inside these boxes is a bit of joy.

However, due to some shipping difficulties, my present didn't arrive until three days after Christmas.

Needless to say, I was surprised to open the box find a stock K3/100!

She had the presence of mind to get me the unassembled version, instead of the factory-built. While it's a bit more work, I really enjoyed assembling my K2/100, so I figured I'd get some enjoyment out of this.

Biggest problem, though, was where to build. The Micro-Shack is, well, micro. The entire desk surface is barely 5 by 2 1/2 feet. Most of my tools are still down at the QTH in Gwinnett county. Nevertheless, I figured I'd do what I could with what I had, rather than wait a week or so until I could get to my big workbench.

K3/100 goes together on the world's smallest workbench.

As it stands, I had enough room for the job.

It was already late in the day when I got started, so I decided to forgo the Elecraft-recommended inventory. I've built a number of their kits, and I can't say I've ever been shorted a part. Indeed, they always seemed to ship me extras. In retrospect, this wasn't such a bad idea. If you've never built a kit before, inventorying the kit is a great idea.

Since all the boards are pre-assembled, there's no need to warm up a soldering iron. However, there's plenty of mechanical assembly. I thought it was weird that Elecraft has one put the left and right side panels on, only to have you remove them a few steps later.

I had never really seen the insides of a K3 before, and the pictures don't do it justice. In some ways, it really is like an over-sized K2, with the mechanical rigidity coming from the main board. Once you get the casing on, it's really pretty rugged.

Initial power-on successful.

Assembly was pretty much a joy, except for the limited work space. I had a great sigh of relief the first time I powered the K3 on, since it was clear I hadn't messed it up somehow.

I did have one item I did mess up during assembly. It was the KREF3 board. The instructions indicate you are to cut off any leads that might be longer than certain components on the board. I found a through-hole connector lead that I thought was a bit long, and when I tried to trim it, I also managed to bust capacitor C2 off the board. Fortunately, this component isn't used for anything critical. My guess is that it is used for FM only, but there's not really any documentation as to what the TXMIXFIL signal does. I've already contacted Elecraft support as to how they suggest we fix this.

After about six hours of work, my K3/100 was all assembled and calibrated. I only ran into one issue -- I had no dummy load to perform the Transmitter Gain Calibration. I ended up doing the calibration manually using the antennas I have on hand. Not perfect perhaps, but it will do until I can get my hands on a dummy load.

K3/100 assembled and operational.

Now that it's assembled, I have to go an order the other options I need. Unfortunately, the KAT100 doesn't work with the K3. I went ahead and used some of that "K3 Fund" money and ordered the KAT3 antenna tuner as well as the KXV3A transverter adapter. That should do me for quite some time.

Now, I need to sit back and read this thick operating manual so I know how to use my new rig! See you on the bands.

Making Noise

When calibrating the Elecraft K2 crystal filters, a noise generator becomes an invaluable tool. Using one results in clean spectrum traces. I had built an RX noise bridge some time ago from an ARRL Handbook -- the original article for that project appeared in the December 1987 issue of QST.

The noise source is a reversed-biased zener diode -- which produces noise of fairly broad bandwidth. A two transistor amplifier chain feeds the trifilar winding of the noise bridge transformer.

This circuit had the unusual addition of an oscillator to switch the noise off and on at about 1 kHz. This made for the very disconcerting effect of hearing 1 kHz tone no matter where the receiver was tuned. The noise bridge was intended to help me tune a manual antenna tuner without having to transmit, but it didn't work as expected.

The oscillator was useless for tuning the filters of the K2. I modified the circuit by removing the NE555 chip and jumpering pin 3 to pin 4, effectively supplying power to the zener all the time. (In fact, this very circuit appeared in the March 2002 issue of QST, and was also promoted by Tom, N0SS as a noise generator on his site)

However, the two transistor amplifier didn't seem to provide much output -- I got the strongest noise signal when attaching to the base of the first transistor. Perhaps this is why the tuner-tuner project didn't work so well. I suspect there's something wrong with one of the junkbox transistors I used.

My main problem was that I needed to jury-rig it with jumpers every time I wanted to use it. What I wanted was a dedicated noise generator -- which isn't much more than a zener diode noise source, followed by a broadband amplifier.

Since I didn't have much luck with the two-transistor amplifier, I thought of using something different. Mark, WA3YNO suggested using an LM703 as the amplifier. I had a better idea -- why not use a MMIC amplifier? The MMIC offers high gain, unconditional stability and an impedance near 50 ohms. I'm surprised we don't see more MMICs in amateur designs. I had a few MSA 0885's in my junkbox. Perfect.

Better engineers than I have already figured this out. If you look at the Elecraft N-gen, this is exactly what they did. So I build basically the same circuit on a piece of perfboard. The result -- S9+20 dB worth of noise. This circuit is so simple it's hard for it not to work right away. A switch is handy because the circuit draws over 12 mA from a 9 volt battery, so it won't last terribly long if you leave it on.

This noise generator works great. Make sure you don't ever transmit into it, though, as just a little bit of power will destroy the MMIC in short order. I generally use the Rx Antenna jack on the K2 (since I have the 160m module), so there's no chance of transmitting into the noise generator. Just make sure you switch the Rx Antenna off on that band, or you'll be wondering where your signals went. (Don't ask me how I know this)

Configuring Elecraft K2 Crystal Filters

The Elecraft K2 has a number of unique features, one is the variable bandpass filter which can be programmed for different bandwidths. This leads to a very flexible design for CW or RTTY, but requires a bit of configuration work. For each mode, you can select up to four crystal filter configurations (FL1-FL4), including using the KSB2 module filter (OP1). The first filter configuration (FL1) is also used when transmitting.

Typical configurations use OP1 in the FL1 position. For CW, I opted to use 1000 Hz in FL1, then with the progression of 400 Hz, 160 Hz and OP1 for FL4. The 1000 Hz setting is about the widest setting that still has smooth sides from a single peak. Much wider, and one side starts to get a pronounced "hump". This filter is good for general tuning on an active band. 400 Hz is great for crowded contest conditions. I rarely use 160 Hz, but it is useful for digging out weak ones. I centered these on 600 Hz. The OP1 filter is good for tuning a quiet band, but it is harder to zero-beat stations.

SSB and RTTY require OP1 in the FL1 position. On SSB, I used 1.8 and 1.6 kHz settings for FL2 and 3, but I rarely use them, the asymmetric filter makes for harsh listening.

RTTY necessitated a minor modification. During the A to B modifications, I added 47 and 100 pF caps across C174 and C173, respectively for the BFO stability mod. I changed these to 56 and 120 pF. However, this did not shift the BFO frequencies at all. I added 3 pF from the X3/X4 junction to ground and this lowered the lower BFO frequency 800 Hz without affecting the upper frequency. The wider BFO frequency range is necessary to accommodate the higher tones use on RTTY.

I originally picked the "low tone" frequencies of 1275 / 1445 Hz for RTTY. However, the resulting center frequency of 1360 Hz, can't be used with the KDSP2 filters. The KDSP2 filters can only be set to multiples of 100 Hz. So, I switched to frequencies of 1415 / 1585 Hz, nicely centered around 1500 Hz.

For the rest of the RTTY filters, I used 1000 Hz, 500 Hz and 300 Hz. 1000 Hz is good for general listening, 500 Hz is pretty much single-signal, and 300 Hz clips the edges of the signal passband, but is useful for trying to dig out stations with surrounding QRM.

Align the filters using the standard procedures for the CW and SSB filters, the RTTY filters are pretty much the same as the CW filters, except the center frequency is different. Instead of Spectrogram, I use CocoaModem on the Mac. If you go to the Config window, CocoaModem has a nice spectrum display. You can set the RTTY frequencies as markers in the window. You can see the picture at the top of the article shows CocoaModem displaying the 400 Hz filter spectrum, with markers at 550 and 650 Hz.

There's two things to know about setting the K2 filters. First, there's a limit to the resolution of the DAC used to tune the BFO -- it may not be possible to get the BFO of each filter exactly on the right frequency. The net result is that the frequency of a desired signal may shift slightly when moving from filter to filter. Second, it's important to match up both sidebands so they look and sound the same. Because of the VCO design, the K2 inverts signals above 20 MHz, so the filter used for LSB is used for USB above 20 MHz and vice versa. (The K2 firmware takes care of this automatically)

When setting up the filters, it's helpful to try to flip between opposite sidebands (using the CW RV button), and check that both sidebands sound about the same. Bumping the BFO setting by one tick can often be helpful in getting them to match.

Similarly, walk through the filter settings (using the XFIL button) to monitor any shift in the received signal. Moving the BFO setting one tick can help in some cases, but because of the DAC resolution it won't be completely perfect. With a little patience, it isn't hard to get the filters as close as possible.

The filters in the K2 offer excellent performance in an inexpensive radio, it just takes a bit of care to configure them.