On 8/28/21 6:30 AM, Black Michael via wsjt-devel wrote: > 50Hz channels won't work due to skirts on the 50hz spread, rig > frequency inaccuracies and sound card non-linearity effects (e.g. > 2000Hz is not the same real freq offset as 200Hz). 60 Hz would be a > better channel width or 55Hz might work. > > Then you have operators whose bandwidth is limited but that could be > accounted for by the waterfall window width.
Yeah, this scheme could easily be limited by the worst receiver. The same is true in so many other situations: satellites, packet radio, ... That makes me think the ultimate solution, as I said in my last message, will be a separate software receive SDR. They're cheap and getting better all the time. You do need some sort of T/R switch. I have one from MFJ that works pretty well. It just parallels the transceiver with the SDR receiver port, and because it's HF you can parallel a bunch of receivers on the port with T connectors; no fancy splitter required. When the transmitter keys up, a relay disconnects the SDR receivers. Seems to do it fast enough; I've never had a problem with it. But I know you can't expect every wsjtx user to do this, so we still need to solve the problem of existing hardware. > > For all those think about this as a solution you should test your > accuracy (and few operators do this unless you are an FMT nut). Once > you have frequency calibrated you can determine non-linearity by > placing your frequency source at different offsets in the audio > spectrum and seeing if the freq error remains constant. > > FLDigi has a nice FMT mode where you can measure WWV or a local AM > radio station to calibrate your rig. FLDigi can also determine the > linearity of your sound card. WSJT-X has it's own calibration but > FLDigi's is super-accurate and with proper equipment can measure WWV > to less than 50mHz. Yes, and modern radios seem pretty stable even if they're uncalibrated, so you'd only have to do this once. I do recommend caution on using AM broadcast carriers as references. The FCC only requires them to have +/- 20 Hz frequency accuracy, and from my observations they're all over that range. Depending on how you use them their error could be multiplied at HF. WWV is of course much better but they're limited by ionospheric propagation. Still the easiest reference to use at HF. If your HF transceiver can also do VHF or UHF, a digital TV station's pilot can be useful. But again the FCC rules aren't much help; they basically say your pilot can be anywhere as long as you stay within a channel mask. I've looked at all the pilots visible here in San Diego, and after repacking there is usually more than one per channel. Some stations (e.g, KFMB channel 8 in San Diego) are within 1 Hz of the nominal 309440.559 Hz above the lower channel edge, so they obviously have Cs/Rb references or GPSDOs, while others are 10-20 Hz off. Also, some TV transmitters use a different offset (328843.631957 Hz) to minimize co-channel interference. (Example: KPBS in San Diego.) Your best bet BY FAR is to get a GPS disciplined oscillator (GPSDO), which have become remarkably cheap lately. I got one from China on eBay for $90 and it works great. If your radio has a 10 MHz reference input (as an increasing number of SDR front ends do), you're all set. > > Might be interesting to try just rounding offset frequencies to the > nearest "channel" in the next release candidate series with an > "Enable" checkbox and see how well it works. > That's a great idea. My pseudo-randomization idea could be another option since it would be backward compatible with existing receivers. I'm never sure how to set the transmit frequency option anyway... 73, Phil
_______________________________________________ wsjt-devel mailing list wsjt-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/wsjt-devel
