Now let's see if I can't figure out what makes this thing tick before I start building it...
KX1 Circuit Description
The KX1 manual has a fairly simple description of the signal path, excerpted below in italics, along with my notes, observations, opinions, and other lies. To better understand the KX1 I compare it to my last (and first!) QRP kit, the Heathkit HW-9, as they are very similar in spirit if not design.
Receiver: The receiver is a single-conversion superhet, using down-conversion to a low intermediate frequency (I.F.) of about 4.915 MHz. Down-conversion minimizes complexity and receive-chain noise, while the low I.F. allows adequate CW selectivity with a variable-passband, 3-pole crystal filter. The use of active mixers keeps current consumption low, compatible with portable operation.The single-conversion system is common to the HW-9, and that's prety much where the similarities end. The HW-9 uses a bank of switched crystal-controlled HFO (high frequency oscillator) circuits, each producing a fixed signal which is then combined with a 5.9993 to 5.7495 MHz VFO signal at the 1st mixer (Q107, an MFE131 dual gate MOSFET) to produce a premix signal for the second mixer (U401, a double-balanced mixer) - i.e., for 80m, the crystal oscillator output is 18.330 MHz; 18.330 MHz minus 5.993 MHz = 12.3307 MHz premix signal. The desired RF signal is subtracted from the premix signal and filtered by the 8.830 crystal filter (FL301) to produce the fixed 8.83 MHz IF - for example, 12.3307 MHz minus 3.500 MHz = 8.3307 MHz.
The KX1 is obviously a much more modern design. It replaces all of the HW-9's HFO, VFO and 1st mixer circuitry with a single AD9834 DDS IC to synthesize the injection signal for a single mixer (an NE602). Elecraft says:
DDS VFO: The VFO is based on a low-power DDS (direct digital synthesis) IC. A crystal oscillator provides the reference signal for the DDS, ensuring excellent frequency stability over a wide temperature range. While a DDS-based VFO does not offer the same signal purity as an L-C VFO or PLL synthesizer, it is a good choice for a portable station that will most often be used by a single operator well removed from strong nearby stations. It is also extremely frequency-agile, allowing coverage of both ham bands and SWL bands. The unit chosen (AD9834) requires a minimum of components and draws only 5 to 8 milliamps. The DDS output is filtered by low-pass filter L4/L5/C50-52. In order to provide adequate roll-off of non-harmonic spurs about 14 MHz when operating at 40 meters, the filter's knee is set just above the 20-m band edge. Because of this, LSB receive mode on 20 meters is less sensitive; LSB on this band requires a mixer injection frequency of 4.9+14 = 18.9 MHz, which is significantly attenuated by the DDS filter.
I'm a bit confused about the signal purity comment as it pertains to PLL synthesizers - I always though a big advantage of DDS over PLL designs was the lack of phase noise, but maybe they speak of other issues. For certain, nothing beats a crystal-controlled VFO (the big selling point of the Ten-Tec Omni VI, as I recall). Nevertheless, the DDS advantage is obvious: a single chip in the KX1 virtually replaces the entire oscillator circuit board found in the HW-9, the sole purpose of which is simply to generate an injection signal for a double-balanced mixer.
Analog Devices says, "The AD9834 is a 75 MHz low power DDS device capable of producing high performance sine and triangular outputs. It also has an on-board comparator that allows a square wave to be produced for clock generation. Consuming only 20 mW of power at 3 V makes the AD9834 an ideal candidate for power-sensitive applications."
KX1 designer Wayne Burdick chose the AD9834 specifically because of it's frugality. " The most important design decision was to use a DDS VFO. This would eliminate a number of parts, including the transmit mixer and its crystal oscillator. While it wouldn't provide the high spurious-free dynamic range of an L-C VFO, it would be very stable over a wide temperature range, and also frequency-agile, allowing full coverage of 40, 30, and 20 meters as well as nearby SWL bands. Other designers had used DDS VFOs in QRP rigs with success, notably Dave Benson (NN1G) in his DSW series. But I'd been holding out for a DDS chip with much lower current drain. Luckily, one appeared: the Analog Devices AD9834, which draws just 5 to 8 mA."
Transmitter: On transmit, the DDS outputs the actual carrier frequency, so no transmit mixer stage is required. Q1, Q4 and Q5 form a 3-stage buffer for the DDS signal. Maximum power output from the final stage (Q6) is about 4 to 5 watts, depending on the supply voltage. Q7, in the receive band-pass filter, limits the signal voltage that can reach the receive mixer when the rig is in transmit mode.
A big departure from the HW-9, the KX1's transmitted signal is completely synthesized by the AD9834 chip. Exactly how this happens is a beyond my limited understanding at this time, except that the DDS is keyed by the MCU which (I assume) handles all of the built-in keyer functions, receiver muting, RIT, etc. The final amplifier of the KX1 is a single device, a 2SC2166 NPN transistor (Q6) in common-emitter configuration.
By comparison, the HW-9 transmitter and receiver sections share much of the same signal chain - the HFO, VFO and BFO sections, 1st and 2nd mixer, bandpass and low pass filters. In addition, the HW9 uses two devices (Q405/Q406, both MRF237 NPN-type power transistors).
Microcontroller: A low-power microcontroller (U1, MCU) is used to control the transceiver and handle user interface elements, such as the display and switches. The MCU communicates with the optional KXAT1 ATU on the VFWD/DATA line at about a 2 kHz data rate. U1 runs at just below 4 MHz to avoid band-edge spurs.
Not much I can add here, and there's no comparison to the HW9 here as it is a completely analog rig.
The KXAT1 provides SWR and power information for the KX1 display in TUNE mode. During normal keying, it provides an accurate indication of power output (1 bar per 0.5 watts). Without the ATU installed, the KX1 displays only a qualitative power output indication.
I built the optional manual antenna tuner and SWR bridge along with my HW-9 but don't think I ever used them. They are small units but certainly not as small as the KXAT1 internal board. It uses latching relays so there is no power drain (except while tuning), and according to the QST review it will match a random wire. It also remains in line during receive and is supposed to peak up incoming signals somewhat. I'm sure any sort of filtering will help since the KX1 has no front-end bandpass filters at all, just a low-pass filter.