Best AGC for digital modes?

Buck
Posts: 66
Joined: Sun Apr 09, 2017 3:57 pm

Best AGC for digital modes?

Postby Buck » Wed Jun 12, 2019 4:22 pm

Is there a consensus for the best AGC settings for digital modes? A recent presentation by the Contesting University recommended off if possible, or slow. Thanks for opinions.

Buck ko0y
User avatar
DL2XY
Posts: 116
Joined: Sun Jul 30, 2017 9:47 am

Re: Best AGC for digital modes?

Postby DL2XY » Wed Jun 12, 2019 8:03 pm

AGC Gain as low as possible, as high as necesarry. Any time. Any Mode.

Just drag the green line on peak of the strongest Signal in passband.

Walter [DL2XY]
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Wed Jun 12, 2019 8:15 pm

I've tried fast, and I've tried slow, and seen little difference either way. In a fit of indecision, I've settled on medium. :D

I suspect that it doesn't make a lot of difference unless you are riding the ragged edge of FT8 or something like that.
K9RX
Posts: 414
Joined: Fri Apr 14, 2017 3:47 pm

Re: Best AGC for digital modes?

Postby K9RX » Thu Jun 13, 2019 1:12 pm

AGC gain: the only caveat is the gain line shouldn't be set too high (on the screen, low value). If grossly high you will get vertical lines printing on the WSJT waterfall ... with a bit less than that you can miss weaker stations... I set the line just above the noise level (15 - 20 db above it)

AGC mode: I run it on FAST. Having it set to OFF is a far cry from SLOW. I'd think they'd recommend off or fast.

That stated as Scott said: I've experimented with it quite a bit and other than the above caveats found there's little obvious interaction.
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Thu Jun 13, 2019 3:28 pm

I find that I have to keep the AGC gain line way high on the screen (gain value of 40 or so) in order to get 30db showing in Ft8 software on a quiet frequency. If I set it any higher in gain terms the audio level is too high. Am I missing something ?
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Fri Jun 14, 2019 5:59 pm

Tony EI7BMB wrote:I find that I have to keep the AGC gain line way high on the screen (gain value of 40 or so) in order to get 30db showing in Ft8 software on a quiet frequency. If I set it any higher in gain terms the audio level is too high. Am I missing something ?

Tony,

It is important to keep your AGC reference line within 10dB of noise floor as shown on the spectral display. If you use that to adjust audio levels into the software (assumed to be WSJT-X?) then you are going to miss out on decoding the weaker stations.

For example, I just tuned in 20M FT8 using WSJT-X. I have a DANL of about -136. I put the green line at -128 which resulted in an AGC gain of 107 and a quiet band With those settings I was reliably seeing decode SNRs from +9dB down to -24dB. And with those settings, and my audio gain settings (which I'll get to in a minute), I'm seeing a quiet band audio level of about 68dB and about 73dB when signals are being received.

Now, when I move the AGC gain by 40dB, to a setting of 67, my quiet band audio level is about 30dB, and when signals are being received this goes to about 53dB. This is a huge improvement in audio dynamic range, and it looks great on the WSJT "Wide Graph", but it has a negative effect on decodes. Under these conditions I see decodes from +9dB down to -20dB. I'm not decoding the smaller signals.

With a high AGC gain setting of course there is going to be compression in the passband. But this appears to be OK from a reported SNR perspective. However, I did not make a scientific assessment of the quantity of signals decoded. It is possible that I actually decoded more signals with the lower AGC gain setting. Or less. I don't know. By eye it looked about the same (there are a lot of signals on 20M right now). Perhaps that's something you might look at.

All that said, if you choose to use a higher AGC gain setting and it's driving your audio levels into the red in WSJT-X, simply reduce your audio gain at some convenient place. You can do this with the RX1 AF control, or the VAC RX Gain control, or in Voicemeeter if that's what you are using, etc. I always run a gain reduction of -10dB, myself.

73!

Scot
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Fri Jun 14, 2019 6:45 pm

Scott thanks so much for the very informative reply its most helpful . I think its pretty obvious I was way wide of the mark with my set up. In order to get near 30 on the audio slider on JTDX software I've backed off the VAC1 rx gain to -30 and the RX1 AF ( which I had totally overlooked) to 15. I can now get the AGC gain line down closer to the noise floor on 6m . One thing I notice though is the signal meter is not showing the same value as the grid scale . I wonder is there something else I need to adjust, perhaps the Rx16M LNA value which is currently set at the defualt of 13.0 ? The grid scale is currently showing the noise floor as being145 or thereabouts . Pic of my Thetis screen below.
Image
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Sat Jun 15, 2019 2:46 am

Tony,

I don't know what a normal noise floor looks like on 6M because I never use it. I'm not sure what you mean by "RX1 6M LNA value" because there is no such thing. The 6M LNA gain is not adjustable, you can only turn it on and off.

I can tell you that something is badly wrong with your audio setup somewhere. -30dB VAC gain and a value of RX1 AF of 15 is essentially no audio at all. This is the equivalent of putting a total of 115dB of attenuation between the Thetis receiver output and JTDX. This problem needs to be tracked down.

- What virtual audio cable software are you using?
- How does everything act when you are on another band, say 20M?
- How does everything act when you use WSJT-X instead of JTDX?

Finally, see this thread: https://apache-labs.com/community/viewtopic.php?f=6&t=2463

73,

Scott
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Sat Jun 15, 2019 7:31 am

Hi Scott , the Rx1 6m LNA value is adjustable on the Calibration tab. I agree something is badly wrong with these levels.

I am using voicemeeter Banana set to Zero level on all sliders except the mic is which is muted.

20m looks the same visually but the meter shows about 20dbm less

Wsjt X acts in the same way.

Thanks for the link I shall study it.
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Sat Jun 15, 2019 12:10 pm

Tony,

The link will answer your question regarding how noise power is measured by the spectral display and the S-meter, but will not solve the audio problem.

The calibration value is just that, a calibration value. It tells the software what the LNA gain IS, it does NOT let you CHANGE the LNA gain. I do not recommend changing this value from the default setting.

Do this experiment:

1. Set RX1 AF to 100, VAC RX Gain to 0, leave all your VMB sliders at 0dB
2. Tune in a quiet part of the band using DIGU and a 3KHz filter.
3. Set the AGC Gain to maximum (green line will be below noise floor).
4. In VMB look at the metering on the Virtual Master channels. You should not see any red segments lit. Push up the gain slider on the master until you just barely see red (the first red bar is +1dB). You should only need a dB or two to get that. That shows you that the level coming out of Thetis is just a hair under 0dB under these conditions. If you get these readings everything is OK. If you don't then something is wrong somewhere.
5. Now select a strong SSB voice signal. Set AGC Gain to a few dB above the noise floor (green line just above the noise floor).
6. Repeat step 4. You should get the same results. Again, if not, then something is wrong between Thetis and VMB.

73,

Scott
K9RX
Posts: 414
Joined: Fri Apr 14, 2017 3:47 pm

Re: Best AGC for digital modes?

Postby K9RX » Sat Jun 15, 2019 1:59 pm

Wow Scott.... seems based on your assessment I'm way off ... I run, HAVE TO RUN my VAC RX gain at -25. And doesn't the RX AF depend on the Master setting as well?

I ask only because I have no problem getting decodes in the -20 through -23 range and occasionally a -24 (although FT8 is speced at I believe -21.5) ... and I run the AGC at about 20 db above the noise floor on the pan. This gives me a value in WSJT of about 40 db. Isn't THAT the indicator of health - the WSJT value? They recommend 30 - I run it typically at 40 ... yet you've said "that is no audio at all".

My DANL on 10M (and 12) is around -145. My S meter reading, with 3Khz bandwidth setting and no signals is -115 with SIGNAL selected and -122 with Signal AVG selected (in both cases the pan has the AVG turned on, not sure if that matters).

Again - if I can tweak more out that would be great. That stated I'm currently #1 in the world for ClubLogs DXCC Challenge at something around 220 countries worked this year on DIG mode. So something must be right.

Gary
K9RX
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Sat Jun 15, 2019 2:45 pm

Gary,

Please go back and re-read my posts. I don't think you are reading them correctly.

K9RX wrote:Wow Scott.... seems based on your assessment I'm way off ... I run, HAVE TO RUN my VAC RX gain at -25.
First of all I would not say you are "way off". And second of all you don't "have to" :) . However, this is in no way the same thing as what Tony is going through, as his need to put RX1 AF at 15 clearly demonstrated something is badly wrong. I'm assuming, and we know what that means ;) , that your RX1 AF is at 100. If my assumption is correct, your VAC gain, combined with your chosen setting for AGC gain, is completely reasonable for achieving the audio levels you desire in WSJT-X. I can and just did demonstrate that to myself on my radio.
And doesn't the RX AF depend on the Master setting as well?
As you might remember from my VAC tutorial, it does not. The Master AF only controls the audio output of the radio hardware. The original design intent was to allow the user to have RX1 AF at 100, or nearly so, for VAC audio clients, yet monitor at a reasonable volume via the radio hardware. As more and more people go "virtual" this design intent no longer works so well.
I ask only because I have no problem getting decodes in the -20 through -23 range and occasionally a -24 (although FT8 is speced at I believe -21.5) ... and I run the AGC at about 20 db above the noise floor on the pan. This gives me a value in WSJT of about 40 db. Isn't THAT the indicator of health - the WSJT value?
That is only a single indicator. There are many other indicators and controls that are just as important, if not more important. For instance, you can achieve the desired audio level into VAC with a wide variety of AGC gain settings. Which AGC gain setting is the best at a given audio level? I don't actually know. My experiments to show Tony that putting the AGC line 40 or 50dB above the noise floor clearly demonstrate that you need more AGC gain than that, that's all. Your setting of 20dB above the DANL is not so different than mine at 10dB above.
They recommend 30 - I run it typically at 40 ... yet you've said "that is no audio at all".
I never said that about that level. I did say it, but it was in reference to Tony's settings. Read again, please.
Again - if I can tweak more out that would be great. That stated I'm currently #1 in the world for ClubLogs DXCC Challenge at something around 220 countries worked this year on DIG mode. So something must be right.
It's great that you are the champ! Congratulations! However I suspect that perseverance, technique, skill, and antennas have way more to do with it than tweaking the last dB out of your AGC and audio settings. Clearly they must be close to optimum, though. In any case, yours and mine are far, far different than Tony's. He's got some sort of audio problem that he needs to find and fix.

73,

Scott
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Sat Jun 15, 2019 3:15 pm

Thanks again Scott, carried out the tests as you suggested and on the SSB signal VMB is well into the red so there does seem to be a an issue between Thetis and VMB . On a side issue may I ask a favour can you tell me what the default value is for 6m LNA in your copy of Thetis please, I copied this over from a power sdr screenshot so I may have changed it at some point ?

Edit: coincidentally the SSB signal I listened to was from an Anan 7000, what are the chances :-)
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Sat Jun 15, 2019 5:00 pm

Tony,

The default is 13dB.

Post screen shots of the following things (use the Attachments feature of the forum editor below the main text entry box):

Thetis VAC settings
Voicemeeter main UI
Voicemeeter > Menu > System Settings Options
WSJT-X or JTDX audio setup
VB Audio Virtual IO Control Panel
VB Audio Virtual Aux IO Control Panel

Also, make sure that in the Windows Sound Control panel, ALL Voicemeeter related devices are set to 100% (0dB) levels (you can right click on the number and change from % to dB as you desire) AND 48KHz. That's ALL Voicemeeter related devices, both playback and recording.

73,

Scott
K9RX
Posts: 414
Joined: Fri Apr 14, 2017 3:47 pm

Re: Best AGC for digital modes?

Postby K9RX » Sat Jun 15, 2019 5:34 pm

A few additional notes :

No, my RX AF is not anywhere near 100! It is typically in the 30's!
-second - indeed I do acknowledge that 40 - 50 db above is far far too high (too low a gain) ... and you WILL lose signals because of it ...
-three: when I said "30 or 40" I was referring to the audio in 'gauge' on WSJT, it, to me, is the indicator of health. Assuming whatever combination puts it in this range one would be "ok" ...maybe and probably not necessarily optimized - but in the range.
-I adjusted my AGC to put the line just above the noise floor - maybe 5db difference and I seem to, seem to be seeing more -24 signals. HOWEVER... there are, at the same time, just coincidentally a good number of strong to very strong stations - so I offer this:

Last this note... I don't think -24's are real. Here's why. Most everyone should be aware using the WJST waterfall that when there are strong stations and/or ones that have poor IMD characteristics they WILL reduce the readability of other signals... depending on the strength and the IMD issue out to as much as several hundred Hz away. Its when these are present that I will see very low values, -22 to -24, even once, -25. I DON'T think these are accurate, I think there is something in the WJST algorithm that is seeing power differently under these circumstances.

Case in point: I do this all the time (as I'm sure others do) but just an example. Earlier today KH0N (Marianna Island in the Pacific) was on 30M. He was decoding as a -22 consistently and there was a station that was about 450 Hz higher than him that was a +13 that was clearly, as seen in the waterfall, impacting the displayed signal for KH0N. So I moved the filter bandwidth indicator down (from the top of course) to JUST include that strong station and lo-n-behold KH0N now is a -10! Yesterday a similar scenario on 12M but this time a guy that was 300Hz lower and one that was 400Hz higher... moved the bottom up - moved the top down -the station went from -23/no decode to -14 - workable (and worked).

Just a thought to ponder. Also included so that people aren't out there trying to seek the elusive "-24" ...

Gary
K9RX
[using the notch also works for these types of efforts - although its real klutzy and needs to be much better/easier to use as an adjustable notch]
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Sun Jun 16, 2019 11:00 am

Hi Scott , please see attachments on this and the next post, limited to 3 per post it seems.
Attachments
Voicemeeter UI.png
Voicemeeter UI.png (468 KiB) Viewed 17144 times
VAC Audio set up.png
VAC Audio set up.png (126.42 KiB) Viewed 17144 times
JTDX Audio set up.png
JTDX Audio set up.png (19.62 KiB) Viewed 17144 times
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Sun Jun 16, 2019 11:01 am

Second attachments post
Attachments
Virtual IO control panel.png
Virtual IO control panel.png (623.17 KiB) Viewed 17144 times
Virtual AUX control panel.png
Virtual AUX control panel.png (625.56 KiB) Viewed 17144 times
Voicemeeter set up.png
Voicemeeter set up.png (312.03 KiB) Viewed 17144 times
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Sun Jun 16, 2019 2:35 pm

Tony,

Your settings look perfect.

Did you check the level settings in the Windows Sound control panel?

Please do this experiment:

1. Set RX1 AF to 100, VAC RX Gain to 0, leave all your VMB sliders at 0dB
2. Tune in a quiet part of the band using DIGU and a 3KHz filter.
3. Set the AGC Gain to maximum (green line will be below noise floor).
4. In VMB look at the metering on the Virtual Master channels. You should not see any red segments lit. Push up the gain slider on the master until you just barely see red (the first red bar is +1dB). You should only need a dB or two to get that. That shows you that the level coming out of Thetis is just a hair under 0dB under these conditions. If you get these readings everything is OK. If you don't then something is wrong somewhere. Look also at the numerical peak readings just above the meter. Again, they should be right around 0dB.
5. Now select a strong SSB voice signal. Set AGC Gain to a few dB above the noise floor (green line just above the noise floor).
6. Repeat step 4. You should get the same results. Again, if not, then something is wrong between Thetis and VMB.

Thanks,

Scott
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Mon Jun 17, 2019 11:47 am

Hi Scott., windows sound panel set to 100. On quiet part of band with AGC at max I need to increase slider to 2.6 to get the first red bar to show . I can then go to a strong signal and get the red to appear again with AGC just above the noise so perhaps I'm making progress ?
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Mon Jun 17, 2019 1:37 pm

Tony EI7BMB wrote:Hi Scott., windows sound panel set to 100. On quiet part of band with AGC at max I need to increase slider to 2.6 to get the first red bar to show . I can then go to a strong signal and get the red to appear again with AGC just above the noise so perhaps I'm making progress ?

If that is with RX1 AF set to 100 and VAC RX gain set to 0, then that sounds pretty normal.

What happens if you put a noise floor into JTDX with all VMB sliders at 0, RX1 AF set to 100, VAC RX gain set to -30dB, and AGC gain set so the green line is 20dB above the displayed noise floor? What audio level do you get in JTDX?
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Mon Jun 17, 2019 1:47 pm

Scott in that scenario JTDX reads 54db on the audio level indicator
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Mon Jun 17, 2019 5:58 pm

Interesting. I don't have JTDX (and have no interest in installing it). Using WSJT-X, under the same conditions I get 39dB into WSJT-X, which 15dB different than you.

At this point I am at a loss to explain where that 15dB difference is coming from.
Joe-W4WT
Posts: 167
Joined: Sun Apr 09, 2017 5:27 pm
Location: Cumming, GA

Re: Best AGC for digital modes?

Postby Joe-W4WT » Mon Jun 17, 2019 6:44 pm

I use VMP and the Behringer 202HD and see almost the same numbers as Tony. I actually see 60db under the same conditions you just described Scott. I think this is just a difference in gain in JTDX vs. WSJT. I frequently decode all the way down to -26db signals and work them with JTDX so I know the setup is fine. WSJT won't decode signals that weak, in my experience. This is why I run JTDX.

I normally set the agc line at the tops of the "noise" and leave it there. Typically it runs at about 105 with my noise level on 6m to the NE. My RX1 level is only 5 and my VAC RX 1 gain is -20. These settings produce a "44" in JTDX's signal level meter. Everything works great at these levels.

Joe W4WT
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Mon Jun 17, 2019 6:53 pm

Scott I'm double checking now with WSJT X and there does seem to be a difference in levels , WSJTX is showing roughly 15db less for the same set up. I'll post on the JTDX list and see if anyone else is seeing the same to rule out a local issue here.

Thanks again for the help Scott
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Mon Jun 17, 2019 6:59 pm

Joe I have pretty much the same set up as you so its reassuring to read you are seeing similar levels to me.
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Mon Jun 17, 2019 7:27 pm

David MM0HVU on the JTDX mailing list has confirmed there is indeed a difference in how JTDX measures

"Indeed the difference shouldbe around that or 12 ish, JTDX measyres input at Peak signal level, WSJT-x is measuring RMS level which will show around this difference"

"also anywhere in the range (colour coded) from around 40-80ish is fine for input as this will leave enough headromm for strong signals I would think, and using AGCc and AGC on radio will also assist and cannot be shown to have any negative side effects form using AGC."
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Tue Jun 18, 2019 1:31 pm

w-u-2-o wrote:Interesting. I don't have JTDX (and have no interest in installing it). Using WSJT-X, under the same conditions I get 39dB into WSJT-X, which 15dB different than you.

At this point I am at a loss to explain where that 15dB difference is coming from.

I think I might have this sorted out.

The peak-to-average-power ratio (PAPR, or "crest factor") of atmospheric noise is not an exact science, as it will vary depending upon atmospheric conditions. However, the rule of thumb on this is a factor of 8, or 9dB (10log(8)).

I am also going to take a wild guess and say that the JTDX audio implementation sums a stereo input. When you sum a stereo input consisting of two identical mono signals, the signal level will increase by 6dB.

6dB + 9dB = 15dB.

Try switching your feed to JTDX to mono, either by switching to mono in PowerSDR VAC, or perhaps by fussing with your settings in Voicemeeter. You might easily get back that 6dB.
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Tue Jun 18, 2019 2:55 pm

Scott mono/stereo makes no difference
User avatar
w-u-2-o
Posts: 5539
Joined: Fri Mar 10, 2017 1:47 pm

Re: Best AGC for digital modes?

Postby w-u-2-o » Tue Jun 18, 2019 3:08 pm

OK, so now it's a 6dB mystery. But I bet that mono is still being converted to stereo somewhere. Maybe even in JTDX.
User avatar
Tony EI7BMB
Posts: 651
Joined: Sun Apr 09, 2017 2:31 pm
Location: Dublin
Contact:

Re: Best AGC for digital modes?

Postby Tony EI7BMB » Tue Jun 18, 2019 3:14 pm

Yeah quite possibly although David above did mention that it measures differently than WSJT X. With the set up I have now Ft8 seems to work really well , even got a new DXCC on 6m yesterday from Iraq :D

Return to “Digital Mode, Rig Control & Logging Software”