WSJT-X FT8 signal images

[N8UR]

I am running an ANAN-7000DLE with PowerSDR mRX 3.4.9 and WSJT-X 2.1.0 using Virtual Audio Cable under Windows 10. Recently I am plagued with spurious images of any moderately strong signal on FT8 (say, 0 dB or better) where I get "ghost" signals and decodes spaced +/- 100 Hz from the main signal. Depending on the strength of the real signal, there can be two to four of these spurs on either side. This is when operating on 6M; I don't have antennas available right now to check on other bands so I don't know whether it's specific to that band or not (my guess is not).

I've messed around with audio level settings, turning the rational resampler on and off, and pretty much everything I can think of but nothing seems to make a difference.

The only thing I've found on the forum relating to this is this thread from 2017:
viewtopic.php?f=6&t=2681&p=4628&hilit=harmonics#p4628
where the problem was said to occur when the noise blanker was turned on. I saw that and went "aha!" but turning NB off didn't change the behaviour here.

Any ideas what might be causing this, and what the cure might be?

Thanks and 73,
John N8UR

[w9mdb]

Do you have a screen snapshot you can share?

[Tony EI7BMB]

I see this on strong signals on 6m also with NB off. I'll try grab a screenshot next time I see one

[w-u-2-o]

Is this problem specific to only FT8, or does it happen with any strong signal on 6M?

Yes, please post a screenshot, and please make sure it shows the entire PowerSDR window so that all your settings are visible.

[K9RX]

I see this quite frequently. It has nothing to do with PSDR. It is the quality of transmitters as they go higher in frequency seeming to reach a crescendo at 6M!

Note further I use the NB (both NB and NB2) on 6M all the time - no impact at all ... HOWEVER if there are strong nearby signals - like CW or RTTY - they can cause all kinds of havoc in which case you need to turn it off (by nearby : it is a function of the offending signal strength and their separation - a 20 over CW signal 30Khz away can screw with the band with the NB on).

Gary
K9RX
have fun on 6! Heard EU, AF, JA, CAR, CA, SA this season

[Tony EI7BMB]

Gary I'd be interested to see your NB settings if you have time . I definitely see issues when I have to engage Nb/NB2. I know your are not supposed to on digital modes but its so noisy here I've no choice at times.

[Tony EI7BMB]

Heres one , although its very strong this also happens on some weaker signals

FT82020-05-23 18.36.12.png (1.04 MiB) Viewed 14972 times

[w-u-2-o]

If you are talking about that tiny spur within 100Hz of the signal that is not an "image". That is a close-in spur. Given that it is -38dBc, it is probably an intermodulation product ("intermod") that is part of the transmitted signal. Intermod's from the average, un-linearized, amateur radio transmitter are typically in the -40dBc range. They can become visible when the signal exceeds the noise floor by more that 40dB.

[Tony EI7BMB]

Perhaps that pic is not a great example Scott but the images I speak of are 100 hz apart usually. On the transmission in the pic you can see this on the jtdx waterfall which I'll post next time i see a clear example.

[K9RX]

Tony, The NB settings are default (although I've changed the ... threshold I think its called (not at radio) to 15 from 32, makes no difference to this subject though) ... I've not had any issues. There's no reason that i'm aware of that you shouldn't use a NB on DIG modes! NR yes, but not a NB. I use it as i have to - and its extremely effective at times. The only impact on FT8 is with it on I decode more stations and all are stronger (because the noise floor is lower).

the signal you show does indeed look like a local - his signal is TERRIBLE on 6M ... I've written to him - politely - he just says its my RX and dismisses it ... anyway his IMD products are terrible and he often has a 2nd decode value. I see the 2nd decode value commonly on 6M. Unless someone finds a problem with PSDR I'm currently convinced this is simply poor TX design at 6M.

Gary
K9RX

[Tony EI7BMB]

Thanks Gary appreciate the help

[N8UR]

Wow, I didn't realize my post would generate this many comments! I think I resolved the problem, though I'm not certain just what did it.

The ghosts i saw were *not* in the PowerSDR panadapter window, but in the WSJT-X window, and were basically duplicates of the strong signal spaced at 100 Hz intervals above and below, the more distant ones becoming weaker.

In the end, I wiped out all the audio settings in both PowerSDR and WSJT-X and things cleared up. I suspect the culprit might have been using a Windows sound interface other than MME; when I reset it was set to WDM-KS (for reasons I don't recall).

Anyway, after that reset, and using the MME driver, I haven't seen the problem recur.

[Tony EI7BMB]

Interesting, I see the ghosts both in power sdr and JTDX waterfall . I use Asio at 96000 sample rate . Must try a few different settings to see if it makes a difference. using NB definitely messes things up on strong signals . Glad you got your particular issue sorted out , food for thought, thanks.

[K9RX]

N8UR: coincidence! I use MME and see them. My strong suspicion is they are real! I had one local station the other day that was literally 210 Khz wide here! He was a +16 ... and had a scalloped signal down more than 100Khz and up the same! I took a snapshot of it to send to him but it turns out he is one of those that doesn't have an email address... about 15 minutes later all of a sudden he was clean. I suspect someone else who knew him called him and said "DUDE - your signal!"

Gary
K9RX

[db8gk]

I do see the same images, spaced closely apart at around 100Hz, from time to time. In rare cases I actually get multiple decodings from the same signal.

However, since this only happens for some stations. I find the theory of TX side problems due to intermodulation to be quite plausible.

[db8gk]

I think the following picture illustrates the phenomenon quite nicely (recorded with SDR Console on 40m).

There are two strong signals visible. The left one has pronounced intermodulation, at least two or three lines in parallel on either side. The right one, even though much louder, has much less pronounced intermodulation copies.

[img]

ft8.png

[/img]

[N8UR]

I am pretty sure this is *not* what some people are seeing in their PowerSDR panadapter windows. I do not see obvious signs of spurs there when my problem is occurring (though I have certainly seen bad transmitted signals there!).

What I am reporting is visible in the WSTJ-X waterfall display and always results in multiple decodes in the WSJT "band activity" text window.

Further use indicates that the problem is probably on the WSJT-X side as restarting that program will normally resolve the problem for a while, but if left running for several hours, the ghosts may reappear.

[Tony EI7BMB]

Maybe try JTDX and see does the same thing happen ?

[vk6cpu]

I had similar issues with Thetis -- see this thread.

viewtopic.php?f=13&t=3203&p=9213#p9192

Under set up setting VAC to FORCE solved the problem.

[G8VR]

Check this blog out:
http://gm4fvm.blogspot.com/2019/11/ft8- ... e.html?m=1

[N8UR]

Nigel, what you described is definitely part of what I was seeing. This afternoon I was able to nail down what's going on, though perhaps not quite why. The spurs, as well as a frequency wobble, are caused by the rational resampler. See the attached shot which shows a stable, unmodulated carrier at about S9 injected into the receiver. At the bottom is the start with the rational resampler turned off. Then I turned it on and you can immediately see:

(a) frequency wobble over 10 or so minutes while the resampler finds the mark and settles down; and

(b) strong 100 Hz sidebands that stop when the resampler stabilizes.

The little noise burst at the top is when I turned the resampler back off.

So I am virtually certain that the "ghost" decodes I am seeing are the result of these spurs caused by the resampler. When things stabilize they go away, but I am guessing that temperature changes, CPU load, or other things could cause the sample rates to intermittently drift, causing a new stabilization cycle and ghosts while that happens.

I haven't messed with the resampler settings but I think my rates are close enough that simply leaving the resampler off might be my best solution.

[vk6cpu]

Hi John
I also had the same issues with the "ghost 100hz" traces too. But since making ths changes to enabling FORCE on the resampling everything has been fine. BEst of luck