RX1 AF level with WSJTX

User avatar
Tony EI7BMB
Posts: 164
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

RX1 AF level with WSJTX

Postby Tony EI7BMB » Mon Jul 09, 2018 9:35 pm

Just curious as to what guys generally set the level of RX1 AF to when using WSJT X for best signal to noise ratio on receive ?
User avatar
w-u-2-o
Posts: 1144
Joined: Fri Mar 10, 2017 1:47 pm

Re: RX1 AF level with WSJTX

Postby w-u-2-o » Tue Jul 10, 2018 1:04 am

Hi Tony, this is a great question, and I am curious what others do as well.

I have not done any scientific testing, but I tend to run the RX gain at -10dB for both WSJT-X and Fldigi. The waterfall displays in both just look better that way to me.
Joe-W4WT
Posts: 52
Joined: Sun Apr 09, 2017 5:27 pm
Location: Cumming, GA

Re: RX1 AF level with WSJTX

Postby Joe-W4WT » Tue Jul 10, 2018 2:52 am

Tony, I'm glad you asked this as I was about to myself once I exhausted searches for the answer.

I use JTDX and I run my AF gain at 5, my VAC RX gain is at -30, and I vary the AGC gain from about -95 to -120 depending on signal strengths present at the time. I have the AGC in Fixed mode. I'm speaking about operating on 6m now. I seldom run FT8 on the HF bands anymore. Is this right? I have no idea. It seems to work. I know if I see a signal that is pretty strong and it decodes twice or three times during the same cycle, seems to be smeared on the waterfall, then I need to reduce the AGC gain. Other than this, it is really all a guess for me.

I hope someone reads these posts and has a good answer for us!

Joe W4WT
User avatar
Conrad_PA5Y
Posts: 79
Joined: Fri Jan 19, 2018 8:11 pm

Re: RX1 AF level with WSJTX

Postby Conrad_PA5Y » Tue Jul 10, 2018 7:50 am

I think that as long as you use a VAC/VMB path almost any combination of gain is absolutely fine provided that the input to the decoder is sensible. What we are interested in is signal to noise ratio. Shifting the levels using the RX1 level and VMB gain is totally benign because the 'volume controls' have sufficient bit depth and cause no measurable distortion.

Windows mixer MUST be avoided, that does not have sufficient bit depth and there will be truncation errors which does affect signals down near the noise floor. I have proven all of the above to my own satisfaction on EME on 50,144 and 432 MHz.

If I use VAC (rarely)I disable the volume control and just use the RX1 out.

I also run with a fixed AGC wherever possible.

73

Conrad PA5Y
User avatar
Tony EI7BMB
Posts: 164
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: RX1 AF level with WSJTX

Postby Tony EI7BMB » Tue Jul 10, 2018 8:18 am

Thanks for the responses guys. Using voicemeeter banana here so have the VAC Rx gain at Zero , RX AF at 100 and I then back off the fixed AGC so I get just above 30db on JTDX receive level indicator on 6m when no signal present . Just guesswork on my part to be honest as if memory serves I previously read that lowering the AGC gave best s/n . Would love to hear more opinions on this.
Last edited by Tony EI7BMB on Tue Jul 10, 2018 11:13 am, edited 1 time in total.
User avatar
w-u-2-o
Posts: 1144
Joined: Fri Mar 10, 2017 1:47 pm

Re: RX1 AF level with WSJTX

Postby w-u-2-o » Tue Jul 10, 2018 10:39 am

I should have added above that I keep all RX audio levels at 100% (0dB). Mainly I do this to make VAC act as a virtual line output. I use VAC exclusively for all operations since my radio is in the basement and the operating location is in a second floor office. I do all of my level control outside of PowerSDR in either Voicemeeter Banana or in the DAW I use (Reaper). Again, I typically adjust (externally) the feed to the digi software down 10dB from that.
User avatar
Tony EI7BMB
Posts: 164
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: RX1 AF level with WSJTX

Postby Tony EI7BMB » Tue Jul 10, 2018 11:13 am

I also use waterfall zero and waterfall gain in JTDX software to get the waterfall looking right.
K9RX
Posts: 196
Joined: Fri Apr 14, 2017 3:47 pm

Re: RX1 AF level with WSJTX

Postby K9RX » Tue Jul 10, 2018 12:24 pm

Joe-W4WT wrote: I know if I see a signal that is pretty strong and it decodes twice or three times during the same cycle, seems to be smeared on the waterfall, then I need to reduce the AGC gain. Other than this, it is really all a guess for me.
Joe W4WT


Joe, Is this true? I always thought those smeared double and sometimes triple decodes were caused by bad TX at the other end ... I've played with settings and never saw a cause-effect relationship to them ... either they're there or they're not there (there with some stations - not there with most others). I've seen very strong signals, in the +15 - +17 range be as clean as can be expected with no smearing ... and others, almost always locals, smear frequently.

I have my AGC set to Fast. Never found any real difference going to FIX. I run the VAC RX at -10.

Gary
K9RX
User avatar
Conrad_PA5Y
Posts: 79
Joined: Fri Jan 19, 2018 8:11 pm

Re: RX1 AF level with WSJTX

Postby Conrad_PA5Y » Tue Jul 10, 2018 2:18 pm

The multiple 'ghost' signals are often caused by a badly adjusted TX chain but not always. On occasion I have seen them caused by aircraft scatter and other multipath effects. Of course non linearity in the RX chain can cause it but if you have a purely digital audio path then I cannot imagine that this will occur. Setting the AGC to fixed really only helps for very weak signals, in the Es season it is more of a hindrance than a help on 6m. Too many very strong locals!

As I said previously, provided the decoding program that you are using is fed with a sensible level then the SNR will be preserved and all will work fine. There are multiple ways of doing this none of which are wrong, you are merely shifting levels mathematically. Providing the calculations have adequate resolution then there is no penalty for this. It is not like optimising the SNR on an ADC input.

Personally I am more inclined to trust the PowerSDR RX gain and VMB volume control rather more than any fader in a decode program but you'd be very hard pushed to see any difference.
User avatar
w-u-2-o
Posts: 1144
Joined: Fri Mar 10, 2017 1:47 pm

Re: RX1 AF level with WSJTX

Postby w-u-2-o » Tue Jul 10, 2018 2:32 pm

I'm not a big DX chaser on FT8. But I do chase weak signals on Olivia in Fldigi. FWIW I find medium AGC to work well for this.
K9RX
Posts: 196
Joined: Fri Apr 14, 2017 3:47 pm

Re: RX1 AF level with WSJTX

Postby K9RX » Tue Jul 10, 2018 2:46 pm

FT8 has proven to be a serious threat to activity in all other modes on the bands... in the past when conditions were just marginal I'd be able to work lots of DX say on 15 meters (using 6/6 stack) ... but now even with signals, MANY of them on FT8 and levels well above +0 there is for all practical purposes no activity on either CW or SSB ... the flip side is I've added about 25 countries to my 6 meter DXCC total this season alone and I have a somewhat crippled 6M antenna ... it is phenomenal at times - I've seen decodes of ONLY DX coming through - 7 DX stations in one sequence .... one day the whole of the middle east was coming through (alas I was unable to work but one of them that day) ... hopefully a new 8 element Innov will help for next season.

Gary
K9RX
User avatar
w-u-2-o
Posts: 1144
Joined: Fri Mar 10, 2017 1:47 pm

Re: RX1 AF level with WSJTX

Postby w-u-2-o » Tue Jul 10, 2018 3:09 pm

JT modes killed off nearly all digi mode traffic in about 6 months. FT8 killed JT modes in under a week. We should either give Joe Taylor a medal or banish him, not sure which ;)

It's funny: I asked Joe to automate JT65. He was horrified by the idea. I guess he changed his mind. We are only a few lines of code away from complete automation. You still have to click to log and start another exchange. I'm surprised nobody has done it yet.

Sorry for the thread drift...
User avatar
Tony EI7BMB
Posts: 164
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: RX1 AF level with WSJTX

Postby Tony EI7BMB » Tue Jul 10, 2018 4:44 pm

Medium AGC, interesting Scott, will try that . 6m dead past two days so might take a while . Last person in log of note is K1JT of all people :-)
K9RX
Posts: 196
Joined: Fri Apr 14, 2017 3:47 pm

Re: RX1 AF level with WSJTX

Postby K9RX » Tue Jul 10, 2018 5:40 pm

sorry to continue the divergence: until someone else posts additional info/questions on this subject let me finish with this ...

Scott, apparently you've not seen the "video" ... I'll need to look for it again - but there is a video of someone in EU that HAS fully automated FT8. It is running 24/7 (if I recall) with zero need for human support/interaction. He's using macro's and the like.

Gary

http://ae5x.blogspot.com/2018/01/video- ... -qsos.html
Joe-W4WT
Posts: 52
Joined: Sun Apr 09, 2017 5:27 pm
Location: Cumming, GA

Re: RX1 AF level with WSJTX

Postby Joe-W4WT » Wed Jul 11, 2018 12:51 am

Hi Gary,

I'll have to check that again to make sure I saw what i think I saw. I know there are instances where there are transmitters that are just not adjusted properly putting out all kinds of garbage. We have one local that sprays crap all over the 2.7khz range on 6m so I know not all "smearing" is agc related. I'm looking at 6m right now but signals are pretty weak, for a change. I'll look tomorrow and see if I can cause a double or triple decode to clear up with agc again and if so, try to capture the screen.

Conrad mentioned airplane scatter could cause the ghost signals. I wonder if that is what I see mostly as I am in the Atlanta area and we have LOTS of planes coming over all the time. I imagine if that is mostly what I'm seeing, it could come and go and I may have thought adjusting the AGC had affected the signal but perhaps didn't. More work to do to see for sure.

Since the consensus seems to be that fixed gain really doesn't offer much, I'll probably go back to fast or medium which I used to run until I read somewhere that "fixed" was the way to go. Conrad is probably right in his comments above in that as long as the correct amount of signal is fed to the decoder, the various combinations of gain controls used don't matter.

Joe W4WT
User avatar
Conrad_PA5Y
Posts: 79
Joined: Fri Jan 19, 2018 8:11 pm

Re: RX1 AF level with WSJTX

Postby Conrad_PA5Y » Wed Jul 11, 2018 8:46 am

If you have a quiet location try the following.

As an experiment go on 6m or 10m late at night when the band has died and use fast or medium AGC. If on 6m Make sure that you have the 6m LNA on. Set your threshold to be just above the noise floor as recommended in the user guides and listen to the band noise with headphones.

Now switch to fixed AGC and adjust the volume to be comfortable and listen to the band noise again in the headphones.

Do you notice any difference in the way the noise floor sounds? You will notice something obvious I am fairly sure. Please report your findings here.

I suspect that the same thing can be observed with a dummy load connected instead of an antennas but I have not checked that yet.

73

Conrad PA5Y

PS Scott I feel the same way about Joe and his bloody FT8 :-)
User avatar
w-u-2-o
Posts: 1144
Joined: Fri Mar 10, 2017 1:47 pm

Re: RX1 AF level with WSJTX

Postby w-u-2-o » Wed Jul 11, 2018 1:14 pm

K9RX wrote:sorry to continue the divergence: until someone else posts additional info/questions on this subject let me finish with this ...

Scott, apparently you've not seen the "video" ... I'll need to look for it again - but there is a video of someone in EU that HAS fully automated FT8. It is running 24/7 (if I recall) with zero need for human support/interaction. He's using macro's and the like.

Gary

http://ae5x.blogspot.com/2018/01/video- ... -qsos.html

Gary--that's brilliant! Thanks for the link. I'd like to see his code for how he stops failed QSOs, though.

I saw in the video comments that perhaps JTDX includes this level of automation, now? Is that correct?

Obligatory receive quality note: I will try your experiment with AGC, Conrad.

73,

Scott
User avatar
Tony EI7BMB
Posts: 164
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: RX1 AF level with WSJTX

Postby Tony EI7BMB » Wed Jul 11, 2018 1:17 pm

Scott I have the very latest "bleeding edge" JTDX and it does not have full automation so far. Given the amount of bad local TX's I would hate it if it did :-)
Joe-W4WT
Posts: 52
Joined: Sun Apr 09, 2017 5:27 pm
Location: Cumming, GA

Re: RX1 AF level with WSJTX

Postby Joe-W4WT » Sat Jul 14, 2018 1:42 am

Hi Conrad,

I did your noise test and what I hear is the background noise (this was on 6m) is quite uniform with Fixed AGC. If I switch to Med or Fast and adjust the AGC to the top of the noise, the background noise sounds "choppy" is the best way I can describe it. Interesting. Why is this?

I also played with some weak double hop signals out to the west coast switching between Fixed AGC and Med AGC and with weak signals I sometimes saw stronger signals for the same stations that I did with Med AGC. Of course there is 30 seconds between seeing the signal with each AGC setting so it is difficult to sort out band condition changes from actual AGC differences.

I haven't see any "ghosting" in the last few days to play with settings on that but I'm sure I will soon.

Joe W4WT
User avatar
Conrad_PA5Y
Posts: 79
Joined: Fri Jan 19, 2018 8:11 pm

Re: RX1 AF level with WSJTX

Postby Conrad_PA5Y » Sat Jul 14, 2018 8:02 am

Hello Joe. EXACTLY!

I believe that It happens because the AGC is unsure what to do at very low levels, there is some kind of toggling going on. The only way to get rid of these artifacts is by setting the threshold so high that you will lose sensitivity. So for weak signals it is imperative that fixed gain is used, What you hear has a profound effect on weak signals near the noise floor. The decoder has to deal with the noise floor changing and it definitely has an effect.

So for weak signals use fixed AGC. Once the band noise is so high that the noise floor sounds even with AGC hen of course use it. On 6m this constantly evolves as conditions change.

73

Conrad PA5Y
User avatar
w-u-2-o
Posts: 1144
Joined: Fri Mar 10, 2017 1:47 pm

Re: RX1 AF level with WSJTX

Postby w-u-2-o » Sat Jul 14, 2018 1:40 pm

I could not duplicate the "choppy" sound on the AGC experiment. On a quiet band, or with the dummy load connected, it sounds identical to me in Medium or Fixed AGC modes on either 80 or 10M. I did not try 6M (I don't work 6M).

I might try this experiment again and feed the audio into a spectrograph type instrument (courtesy of my DAW software) if I get some more time, just to see if I can visually detect any spectral or temporal differences that I cannot hear.

This was on an 8000, BTW.
User avatar
Conrad_PA5Y
Posts: 79
Joined: Fri Jan 19, 2018 8:11 pm

Re: RX1 AF level with WSJTX

Postby Conrad_PA5Y » Sat Jul 14, 2018 4:00 pm

You will have no problem hearing it if it is there. The only instruments needed are attached to to the side of your head. It is glaringly obvious. Maybe it does not happen with the 8000. I have observed it on two ANAN-100Ds and an ANAN-10 so cannot comment.

73

Conrad PA5Y
Joe-W4WT
Posts: 52
Joined: Sun Apr 09, 2017 5:27 pm
Location: Cumming, GA

Re: RX1 AF level with WSJTX

Postby Joe-W4WT » Sat Jul 14, 2018 4:41 pm

Hi Conrad,

It happens with the 8000. That is the board I use. Perhaps it is more difficult to hear on HF due to the higher noise thresholds.

Joe W4WT
User avatar
Conrad_PA5Y
Posts: 79
Joined: Fri Jan 19, 2018 8:11 pm

Re: RX1 AF level with WSJTX

Postby Conrad_PA5Y » Sat Jul 14, 2018 4:59 pm

OK Joe thanks. I think that maybe a resistor does not work, I must have been mistaken there. Perhaps what is needed is a variable noise floor that will occur when an antenna is connected. Can you hear it with a simple wire antenna on 10m?

Regards

Conrad
User avatar
DL2XY
Posts: 24
Joined: Sun Jul 30, 2017 9:47 am

Re: RX1 AF level with WSJTX

Postby DL2XY » Sat Jul 14, 2018 7:30 pm

On pure white noise there is virtually no difference to detect.

With slightly fluctuating band noise you can reproduce the effect by setting the AGC-Level within the fluctuation band.
Everytime the level exceedes the threshold the gain is lowered to come up again with decay time constant.
That is perfectly the way an AGC should work.

To get an equal overall level do not rely on the "Gain Line" in spectrum display, use only the numeric gain display over the AGC Gain slider.

Regards Walter

Return to “Digital Mode & Logging Software”