I have been running tests on K3s and K4s to understand the conditions needed to transmit clean FT8 signals. I have previously shared some of my findings with Pete N4ZR and a couple of other folks. Their feedback has been very valuable.
Joe Taylor, K1JT has shown the spectral purity of the FT4 and FT8 protocols (using WSJT-X version 2.1 and later) to be very clean to 80 dB below the peak signal. See https://wsjt.sourceforge.io/FT4_FT8_QEX.pdf (Figure 3 and related discussion on pages 9 and 10) and https://www.w2zq.com/wsjt-x-innards-explained-by-joe-taylor-k1jt/ (slide entitled Spectra: RTTY, FT8, FT4). If the WSJT-X operating instructions are followed (use operating mode "Fake-It" or "Split"), there is exactly one clean signal to be transmitted as discussed in the exchanges in this thread https://groups.io/g/Elecraft-K4/message/25371 I have been endeavoring to ensure that ALL of the FT4 and FT8 spectra purity shown by Joe Taylor can be replicated by my K3 and K4D in the Data A mode - and that the recipe for a clean spectra can be shared with others. Clean FT8 spectra is badly needed, especially in metro areas at 10 meters and above, where many poor quality FT8 signals abound and wipe out many of the weaker stations that other stations are trying to receive. Yes, there are suggestions that folks in the same area all transmit at the same time, and many people try it, but in practice, not everyone is on-board with that suggestion. But, there are also stations on the air that transmit very clean FT8 signals every day, so it can be done! I shared with Pete that neither my K3 or K4 transmitted the cleanest possible FT8 signal using either analog or USB digital inputs and the conventional setup instructions that have been often repeated on this reflector (i.e. set the audio level so there are 4/5 bars of ALC showing). After I beefed up my DC power cable (it is now 2 feet of #10 wire between the K3 and the power supply), I found that setting the Line In level to the point just before the 1st ALC bar lights up (i.e. Line In = 2 in my case) results in a very clean FT8 signal. As Pete reported, there is a slow increase of RF power output level for 10-20 seconds as the ALC circuit compensates for the lower input level, but apparently the ALC loop remains open, hence the clean signal. For my K4, the FT8 signal is clean with no more than 2 bars of ALC. To be clear, the spurious responses I have seen with ALC set to 5 bars has NO measurable IMD responses. There is only the one strong signal present, the spurious signals being discussed here are all less than 50 or 60 dB below the intended carrier level. My goal is to reduce all the spurious responses to 80 dB below the intended carrier level. The discussion that Pete, N4ZR started several days ago got many responses that could be grouped into three categories: 1. "The manufacturer says use 5 bars of ALC, end of story". I don't buy that at all. I have yet to see any manufacturer's DATA on the relationship between their transmit ALC levels (or any other metric) and the in-band spectral purity of an FT8 like signal beyond "meets FCC spurious signal requirements". 2. "I must be mistaken, my equipment is overloaded, etc". In years past, that would have been a fair criticism, it would have taken hundreds of thousands of $$ worth of calibrated test equipment to prove whether anyone's receiver was overloaded or not. With today's high end receivers (the K4 being one of them), the overload point in every gain state is well known (notably confirmed and publicized by NC0I, Rob Sherwood), and overloaded signals trigger a visible alarm on the radio. The K4 (and I assume other high end receivers with similar architecture) has a repeatable performance (amplitude stability) in showing signal levels to the dB on a panadapter. To me, external monitors and external display programs (like Win4K4) help to visualize and collect data better than trying to read the data off the radio's relatively small screen. 3. "Don't use my data, have someone near you with a high end receiver /panadapteror a good spectrum analyzer (and who knows how to use it) look at your signal". I agree, that is a very sound approach. On 6 meters, a station 10-20 miles away, and near line of sight should provide a good signal strength for analysis by a high end receiver/panadapter. It is also possible for any transmitter that more than one problems exists, beyond the ALC issue. Other potential problems include poor power supply regulation, excessive DC voltage drop on transmit, ground loops, or a defective phase locked loop. Those problems should be fixed before changes to the ALC response is attempted. The K3 accepts both analog and digital inputs for Data In, with multiple generations of I/O boards. I have not tested all of these combinations! Ground loops in particular can be challenging to clean up. I used a Signalhound SA-44B spectrum analyzer and a Bird 50 dB signal sampler to analyze my transmitted signal. Using the KPA1500 "TX sample" output can replace the need for the Bird signal sampler with very similar results. Hope this helps 73, Mike, W3IP On Friday, July 21, 2023 at 08:58:09 AM EDT, Joe Subich, W4TV <[email protected]> wrote: Are you sure the "technically-qualified local ham" was not running his tests in SSB instead of DATA A? There are more than a few "technically- qualified" individuals who insist on using SSB instead of DATA_A with WSJTX and that makes all the difference in the ALC display. Again, READ THE K3 OWNERS MANUAL and do what the manufacturer says. 73, ... Joe, W4TV On 7/20/2023 10:33 PM, Pete Smith N4ZR wrote: > OK guys - rather than reply individually, let me try to pull everything > together in one message. > > In a previous effort at FT8 I followed the standard advice to set the > power slider in WSJT-X (on the lower RIGHT side) so that my K-3 showed 4 > bars on the ALC meter, just below the threshold for ALC starting, and > that seemed to work fine. Then a technically-qualified local ham showed > me spectrum analyzer images of a K-3 at 4 bars versus at no bars, and > the difference was startling - the no bars signal was clean and narrow, > while the 4-bars signal looked like some of those SSB sigs we hear > during contests, wide and with many artifacts. I am not going to > identify my source here, but I'm fairly sure he reads this reflector, so > maybe he'll join the discussion > > Anyway, I'm kind of at sea - WSJT-X works fine with no bars - I've > finished 12 and 17M, as well as digital DXCC in the last few weeks, > following his advice. But the hunting typically takes almost 20 seconds > of transmitting - roughly 1 1/2 FT-8 cycles - to reach full power - and > happens every time I change bands, which puts me behind the curve if I > switch bands, see someone I need, and have to spend 1 1/2 transmit > cycles getting up to full power before I can call him effectively. > > 73, Pete N4ZR > ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[email protected] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to [email protected] ______________________________________________________________ Elecraft mailing list Home: http://mailman.qth.net/mailman/listinfo/elecraft Help: http://mailman.qth.net/mmfaq.htm Post: mailto:[email protected] This list hosted by: http://www.qsl.net Please help support this email list: http://www.qsl.net/donate.html Message delivered to [email protected]
