Phase noise was recently discussed on the K3 Yahoo group, and I thought I'd add my two cents. Or maybe three :)
The K3's phase noise at 1 kHz is pretty much state-of-the-art for a DDS (direct digital synthesis) reference driving a wide-frequency- range, low-noise PLL (phase locked loop). We took things a step further by using a very narrow crystal filter after the DDS (about 2.5 kHz), dramatically cleaning up the DDS even before application to the PLL. This forced us to use some pretty hairy math in calculating the PLL divider values, but it was worth the effort. The TS590 (and all currently shipping Flex radios) use a synth subsystem that is quite different from the K3's. They use an unfiltered DDS as their local oscillator, with no following PLL. There are some advantages to this design choice. First, and maybe the most relevant: it's cheaper than a DDS-driven-PLL overall, requiring very few analog parts, essentially no alignment, and far less PCB space. Second, such radios might have slightly lower phase noise at some very close offset--although at such spacings, other factors such as keying bandwidth or IMD typically dominate. Finally, use of a raw DDS allows the VFO to switch frequencies rapidly. Such agility might be useful for some digital modulation schemes. However, that raw DDS VFO comes with a price: its output has many discrete spurs that can, at specific VFO frequencies, cause "ghost" signals to appear. This is due to mixing between the DDS spurs and strong signals appearing anywhere inside the receiver's band-pass filter (many MHz in most receivers, but not the K3--more on that later). This is true even with the 14-bit DDS word size described in the TS590's sales brochure. The usual way to eliminate these wide-band spurs is to use a PLL to clean up the DDS's output. Ironically, that sales brochure I mentioned implies that eliminating the PLL was an advantage. Maybe they were thinking about reduced manufacturing cost, though this wasn't stated explicitly. (BTW, a typical lab receive mixing test done at just one test frequency will not necessarily show this characteristic. To reveal the DDS spurs, you'd need to do such a test at many frequencies, moving the VFO in very small increments. This is because the spurs are the product of multiple digital sampling phenomena; they vary rapidly in frequency and amplitude as the DDS's control word is changed. The lack of such testing and transparency in the industry could explain why mixing spurs are *not* a hot topic of conversation among those considering a radio using a raw DDS VFO. Yet, like real ghosts, the resulting signals could, nonetheless, sneak up on you :) It is certainly a lot more expensive to add a high-performance PLL into the system--just ask my engineering and manufacturing staff. But I guess it depends on what you're trying to optimize. We wanted the K3 to perform extremely well in crowded band conditions, so we went to the trouble to use a DDS-driven-PLL synth. (Or TWO of these synths if you have the KRX3 sub receiver installed.) Flex may have elected to go with raw DDS because of the need for a very agile VFO for SDR applications. Kenwood may have been trying to keep costs low. Both are certainly worthy goals. Actually, we made it even harder on ourselves with the K3. We provide narrow band-pass filters on every ham band, painstakingly aligned at the factory, ensuring that as little out-of-band energy as possible is presented to the mixer in the first place. This makes the synth's job a little easier. Yet nearly all other transceivers these days use "half-octave" band-pass filters that are many times the width of the ham-band segment. They require no alignment, but they open the radio up to more interfering signals. (You can add general-coverage band- pass filters to the K3's main and/or sub receivers, of course, by adding KBPF3 module. This has no effect on the ham-band performance.) Note that like the K3, the KX3 uses a DDS-driven PLL synth. The K3 has an advantage in temperature stability since it uses a separate reference oscillator, but the KX3's phase noise is in the same very low range, as evidenced by Sherwood's numbers. Many other factors besides synth phase noise--including transmit signal purity and receiver AGC behavior--also contribute to performance in crowded conditions. This is why, some time ago, we undertook a major redesign of the K3's AGC subsystem. This resulted in excellent field reports from DXpeditions, etc., regarding the dynamics of within-filter signals. I won't go deeply into the SSB transmit purity issue, which has been adequately described by others. But I will mention that the K3's TX IMD at max power output is as good as or better than that of any other 12-volt-capable transmitter. And if you run at lower power when driving an amp (typically 20-70 W), the IMD numbers are outstanding by any measure. 73, Wayne N6KR ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:Elecraft@mailman.qth.net This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html
