|
Hi Bev.
See my notes in this thread on how to roll back to 016 (or whatever you were running before the update). You'll have to remove the 4 bottom screws (2 on each side) to pop the bottom cover off, then use a small piece of wire to jumper those to connection points like I did in my previous post w/ photo.
Start the process with the USB cord plugged into the QMX and your computer, with the power off on the QMX. Now jumper those two connection points (see photo), plug the power cord in, and the folder will show up (unless something else is amiss) to drop the new 017 into it.
I first rolled back to 016, but that's probably not necessary. You should be able to go from original 017 to new 017 firmware directly I'm guessing.
|
So, got a QSO in the log and the good news is the AGC seems to largely work, particularly on Rx.
However it has introduced a very sharp tick at the start of each dit and dah on Tx.? This is very different to the clunk that was present before and sounds more like a loud spark if that makes sense.? I have turned AGC off and can confirm this artefact goes away...
73 Danny M0SDB
|
Thank you Hans! I have been waiting for AGC to take this out on POTA again!!
|
Thank You, Thank You, Thank You! Hans, you are the best.
Hi all
I have released firmware 1_00_017 at
The most important features are the AGC and S-meter.?
S-meter: it appears in the 3 characters to the right of the mode indicator character, when S-meter is enabled. The 15 vertical bars are worth 6dB each and therefore correspond to levels from S0 to S9 (9 bars) and?on up at S9+6dB, S9+12dB etc to S9+36dB when all three are full solid. The scale is calibrated in real decibels, but the accuracy of S0 relative to an actual signal input depends on the level of the gain setting you set for various bands in the Band Configuration terminal menu.?
AGC: it's a quite complex sub-system and I have documented the concepts and parameters in the 1_00_017 operating manual at , section 5.21 pages 52-57 inclusive. Briefly...?
AGC operates in the decibel domain, not linear. There is an AGC threshold below which AGC does not operate, it's a parameter specified in S-meter units. There's an impulse noise filter whose duration is configurable. The AGC slope is configurable so you can adjust it to your taste, you can have a gentle AGC action where strong signals are louder than weak ones, or an aggressive one where the AGC system makes everything the same volume. Hang time and recovery rate are also configurable.?
The AGC system creates an audio sample delay line, which it analyzes for impulse peaks then applies the calculated AGC gain (attenuation) to the sound samples in the delay line. The action of the AGC is therefore effectively instant; no matter how strong the incoming signal, the time taken for the AGC to react to it and protect your ears is always zero,?no piece of strong signal reaches your ears. The length of the delay line is configurable and needs to be a minimum 1.3ms for CW. SSB has a lower minimum audio frequency so a longer delay line is required; the maximum configuration is 6ms which would be able to handle signals down to 83 Hz. Either way, it's a very small latency.?
The S-meter also has a configurable "AGC action" display; if enabled, the S-meter looks like the Power/SWR meter function (during transmit) with two bars. The upper bar is the S-meter and the lower bar is the AGC action, with a configurable number of dB per pixel. It just helps you to see when AGC is active.?
The S-meter configuration parameters were moved to their own sub-menu under the Display/controls menu, and include an S-meter update interval parameter.?
I created a short demonstration YouTube video? . In this video I used a QCX, with about 150cm of wire attached to the antenna port, to generate a strong CW test signal. The BS170s in the QCX are toasted so this just produces very little actual power output but of course we are in close proximity, so the signal level "received" by the QMX is enormous, approx S9+30dB. In the video you can see heavy clipping when the AGC is off, which would no doubt generate a gigantic headache if you had earphones on; when I switch AGC on the signal is reduced to a comfortable S9 level.?
I also tested the QMX with AGC on 40m at night, tuning around for strong signals with and without AGC and I believe it is highly effective.?
The system is highly configurable, I suspect it may be TOO configurable; I will look forward to your feedback and suggestions on how it may be improved, and in particular it may be possible to remove some of these configurations, or derive or calculate them. The attached screenshot shows the configuration I used for the video. I must say though that it does not appear to be a delicate act, balancing parameters carefully; it seems generally to be very effective and insensitive to the exact settings.?
I do highly recommend though that you read the operating manual pages 52-57 carefully; initially all the parameters will be zero and you need to set it up carefully, before enabling AGC. At the moment there are no default values. So let's see.?
Other than that, I also added an LPF sweep tool (see example attached), which was discussed some time ago here. It "transmits" at 1/4 full power, into a dummy load (DO connect a dummy load!) while sweeping the transmit frequency, and plots the measured power output, after conversion to dB and normalizing the center frequency measurement as 0dB. It's indicative only, but quite interesting to see and educational, perhaps useful for debugging too.?
Plus a couple of bug fixes; change list is: - Bug fix: '5351 synth code for setting TX params could lock up in some circumstances
- New LPF Sweep application in the terminal
- Added S-meter
- Added AGC implementation and supporting configuration menu parameters
- Bug fix: Parameter corruption (incl IQ Mode setting lost) in Digi config menu
So that's my week's work here.?
73 Hans G0UPL
|
On 16/02/2024 15:28, algarveastronomy wrote: Except that the COM access and the virtual directory do not appear any more....... Well, after some sweating, I managed to talk to it after rebooting the PC as well as the QMX. I am back on version 16. I don't know whether I will ever have the nerve to try 17 again! Regards Bev
|
No issues? ? Hans,? .......? ?sweet as a nut!? ? ?:-)? ? ? ??
On Friday, 16 February 2024 at 16:43:49 GMT, algarveastronomy <bmes@...> wrote:
On 16/02/2024 15:28, algarveastronomy wrote:
> Except that the COM access and the virtual directory do not appear any
> more.......
Well, after some sweating, I managed to talk to it after rebooting the
PC as well as the QMX.
I am back on version 16.? I don't know whether I will ever have the
nerve to try 17 again!
Regards
Bev
|
Well I loaded 1_00_017 (final "ripened" version) into my 12V rev1 QMX with a QRP Labs dummy load attached.? Loading of the new version? went fine.? I immediately went to the AGC submenu and started entering the AGC parameters as shown in the short YouTube AGC video by Hans aaand..? suddenly the display just showed a top row of black squares then blank.? I turned off power and tried to power up.? It no longer powers up.? It likely has nothing to do with the new version, and is merely a coincidence, but after all this time, and many power ups and power downs, it decided to fail.? Time to do some debugging.? Perhaps the protection zener on the Vdd switching supply? Thoughts?
-Steve K1RF?
------ Original Message ------
Date 2/16/2024 10:20:14 AM
Subject Re: [QRPLabs] #qmx #firmware 1_00_017 AGC, S-meter, LPF sweep
Hi Roy, all
Profound apologies are hereby offered. I would love to be able to tell you that I just fired the entire development time however, I already fired them last time.?
Embarrassment... OUUUCCHHHH... turns out, of course, that the parameters all default to zero, but that can't be OK because then you get a division by zero when loading the parameters, but of course the, err, developer couldn't have tested that scenario which just so happens to be exactly what you would all experience. ALSO - if the sample length happens to be zero then it keeps incrementing until it wraps around the byte counter, well then that tries to write audio samples into a 256-block delay line when the maximum allocation was 9... OH how horrible I should throw myself on my sword.?
Anyway I fixed it all now, you can download a fresh 1_00_017 and it will work OK now. You need, as I mentioned, to set up AGC parameters as it won't be sensible until you do, but, at least it won't hang up on you when stuff is zero immediately after updating firmware.?
I was able, using the wire jumper, to roll back to 016 Firmware, and my darling QMX appears happy again. It's nice that one only needs to remove the 2 bottom screws on each side of the enclosure to pop the bottom cover off (no need to remove knobs and disassemble otherwise).
Roy - KI0ER
|
S meter works well.
With the LPF sweep I suspect that the chosen output power is too much for the LPF test as it is obvious that half of the sweep goes into the filtered zone.
--
73 Bojan S53DZ
|
Hans, thanks very much!
S-meter and AGC seem to work very well, however, as Danny noted, when transmitting, some artefacts like static can be heard on each dah or dit. With AGC disabled, these are gone.
I configured AGC with the values seen in your screen captures, they seem pretty sane, I only changed "Recovery dB/s" (from 10 to 15) and "Threshold S" (from 8 to 7). Couldn't they be used as defaults after a factory reset? It would reduce the hassle involved after a reset, IMO.
Nice work, as always!
Regards,
YO3GFH
op. Adrian
|
Hello Bojan
With the LPF sweep I suspect that the chosen output power is too much for the LPF test as it is obvious that half of the sweep goes into the filtered zone.
Please could you explain? I didn't understand what you're getting at?
73 Hans G0UPL
|
S-meter and AGC seem to work very well, however, as Danny noted, when transmitting, some artefacts like static can be heard on each dah or dit. With AGC disabled, these are gone.
I'll look into that.?
I configured AGC with the values seen in your screen captures, they seem pretty sane, I only changed "Recovery dB/s" (from 10 to 15) and "Threshold S" (from 8 to 7). Couldn't they be used as defaults after a factory reset? It would reduce the hassle involved after a reset, IMO.
Yes I definitely want to set up some reset defaults. Once it becomes clear with some use, what values are typically appropriate.?
The values I used in the video are probably a bit extreme. I was tending to use extreme values because it makes it easier to see what's going on and make sure the AGC logic is working properly. Also easier to see in a demo.
73 Hans G0UPL
|
On Fri, Feb 16, 2024 at 08:15 PM, Hans Summers wrote:
Hello Bojan
?
With the LPF sweep I suspect that the chosen output power is too much for the LPF test as it is obvious that half of the sweep goes into the filtered zone.
?
Please could you explain? I didn't understand what you're getting at?
?
?
Hi Hans, well, perhaps I did not explain my observations well. When doing the SWR scans we transmit a preset portion of RF power into a load/antenna. In my case it is 45%. And the scan stops before the LPF begins to attenuate. Here, on the other hand, when doing the LPF scan we transmit over 1W of RF power thru the LPF into a load but up to the frequencies which are well beyond the cut-off of the LPF. So, most of the scan time we get great VSWR. I get quite a current increase (up to the current limit) when that happens. I feel the 1/4 of full RF power is too much for such a measurement. ? -- 73 Bojan S53DZ
|
Hello Bojan ? Hi Hans,
well, perhaps I did not explain my observations well.
When doing the SWR scans we transmit a preset portion of RF power into a load/antenna. In my case it is 45%. And the scan stops before the LPF begins to attenuate. Here, on the other hand, when doing the LPF scan we transmit over 1W of RF power thru the LPF into a load but up to the frequencies which are well beyond the cut-off of the LPF. So, most of the scan time we get great VSWR. I get quite a current increase (up to the current limit) when that happens. I feel the 1/4 of full RF power is too much for such a measurement.
I'm not sure it's so bad. Problem is, if you use lower power, the noise floor hits you sooner. So the range of the measurement is less. If you use more power, then as you say, VSWR hits. But there is a difference too - the SWR sweep is done into an antenna. The LPF sweep should be done into a more innocent resistive dummy load. I did check the BS170 suffering when sweeping the LPF, done at 25% power output... it seemed harmless to my judgement... of course, I could make it configurable; so the constructor owns his responsibility. Which is also a danger... but I did try at 100% also and still had no fails?or anything, and some extra dB of vertical range of course. But chose the 25% power (50% voltage) setting as it seemed conservative. But maybe I am still too much of a cowboy, I don't know.?
73 Hans G0UPL
|
I got 17 loaded just fine.? Anyone have suggested AGC settings to start with?
73, Bill NZ0T
|
Bill,
I simply used the values from Hans’ video on YouTube.
I put out a CQ on 40m after the upgrade to 1_00_017 and worked Ralph DL1HR who was also using a QMX.
The AGC certainly makes 40m more pleasant to listen to. Thanks, Hans.
73
John
G4YTJ
|
I got 17 loaded just fine.? Anyone have suggested AGC settings to start with?
Adrain?YO3GFH said:
"I configured AGC with the values seen in your screen captures, they seem pretty sane, I only changed "Recovery dB/s" (from 10 to 15) and "Threshold S" (from 8 to 7)."?
I think that is reasonable...?
I remind all that: - I was being a little extreme perhaps in my settings, because I wanted to see what goes on, during my development and testing
- I had long hang time and slow recovery dB/second because again, I wanted to see what was going on, and be able to demonstrate in the video?
- My video test case was also extreme... S9+30dB... to demonstrate the extreme capability of the AGC system to totally delete any possibility of that massive signal reaching the earphones, in this case it instantly applied about 36dB attenuation, and no blip at all is seen on the Audacity software's recording of the audio (which is also streamed over the USB sound card interface concurrently to the audio output). And also to clearly show everything for the video.?
- My 40m noise floor is too high here right now, as I said on the video, my antenna fell down. It doesn't work as well lying on the roof which has a metal frame inside the tiles.?
Easter egg: If you press the D key on the terminal screen, in the main menu, you get my secret "lab", it's my playground which I use during development of my various functional additions, and for debugging, etc. What's in there now is some debug info on the AGC as that's the task I had most recently on the lab bench. Might be interesting, if you are wanting to play with the AGC.?
|
P.S. Attached... "The Lab" Easter egg...?
On Sat, Feb 17, 2024 at 1:25?AM Hans Summers via <hans.summers= [email protected]> wrote: I got 17 loaded just fine.? Anyone have suggested AGC settings to start with?
Adrain?YO3GFH said:
"I configured AGC with the values seen in your screen captures, they seem pretty sane, I only changed "Recovery dB/s" (from 10 to 15) and "Threshold S" (from 8 to 7)."?
I think that is reasonable...?
I remind all that: - I was being a little extreme perhaps in my settings, because I wanted to see what goes on, during my development and testing
- I had long hang time and slow recovery dB/second because again, I wanted to see what was going on, and be able to demonstrate in the video?
- My video test case was also extreme... S9+30dB... to demonstrate the extreme capability of the AGC system to totally delete any possibility of that massive signal reaching the earphones, in this case it instantly applied about 36dB attenuation, and no blip at all is seen on the Audacity software's recording of the audio (which is also streamed over the USB sound card interface concurrently to the audio output). And also to clearly show everything for the video.?
- My 40m noise floor is too high here right now, as I said on the video, my antenna fell down. It doesn't work as well lying on the roof which has a metal frame inside the tiles.?
Easter egg: If you press the D key on the terminal screen, in the main menu, you get my secret "lab", it's my playground which I use during development of my various functional additions, and for debugging, etc. What's in there now is some debug info on the AGC as that's the task I had most recently on the lab bench. Might be interesting, if you are wanting to play with the AGC.?
|
P.P.S. yeah there's a lot of clutter in there. Like my workbench (see attached)... the QMX "Lab" is almost as cluttered as the real one.?
On Sat, Feb 17, 2024 at 1:27?AM Hans Summers via <hans.summers= [email protected]> wrote: P.S. Attached... "The Lab" Easter egg...?
On Sat, Feb 17, 2024 at 1:25?AM Hans Summers via <hans.summers= [email protected]> wrote: I got 17 loaded just fine.? Anyone have suggested AGC settings to start with?
Adrain?YO3GFH said:
"I configured AGC with the values seen in your screen captures, they seem pretty sane, I only changed "Recovery dB/s" (from 10 to 15) and "Threshold S" (from 8 to 7)."?
I think that is reasonable...?
I remind all that: - I was being a little extreme perhaps in my settings, because I wanted to see what goes on, during my development and testing
- I had long hang time and slow recovery dB/second because again, I wanted to see what was going on, and be able to demonstrate in the video?
- My video test case was also extreme... S9+30dB... to demonstrate the extreme capability of the AGC system to totally delete any possibility of that massive signal reaching the earphones, in this case it instantly applied about 36dB attenuation, and no blip at all is seen on the Audacity software's recording of the audio (which is also streamed over the USB sound card interface concurrently to the audio output). And also to clearly show everything for the video.?
- My 40m noise floor is too high here right now, as I said on the video, my antenna fell down. It doesn't work as well lying on the roof which has a metal frame inside the tiles.?
Easter egg: If you press the D key on the terminal screen, in the main menu, you get my secret "lab", it's my playground which I use during development of my various functional additions, and for debugging, etc. What's in there now is some debug info on the AGC as that's the task I had most recently on the lab bench. Might be interesting, if you are wanting to play with the AGC.?
|
I loaded .17 and it is working well on digi with iFTx on the iphone. ?Thanks to Hans ill be trying it out on the next SOTA outing.?
On Friday, February 16, 2024, 3:30 PM, Hans Summers <hans.summers@...> wrote: P.P.S. yeah there's a lot of clutter in there. Like my workbench (see attached)... the QMX "Lab" is almost as cluttered as the real one.? On Sat, Feb 17, 2024 at 1:27?AM Hans Summers via <hans.summers= [email protected]> wrote: P.S. Attached... "The Lab" Easter egg...?
On Sat, Feb 17, 2024 at 1:25?AM Hans Summers via <hans.summers= [email protected]> wrote: I got 17 loaded just fine.? Anyone have suggested AGC settings to start with?
Adrain?YO3GFH said:
"I configured AGC with the values seen in your screen captures, they seem pretty sane, I only changed "Recovery dB/s" (from 10 to 15) and "Threshold S" (from 8 to 7)."?
I think that is reasonable...?
I remind all that: - I was being a little extreme perhaps in my settings, because I wanted to see what goes on, during my development and testing
- I had long hang time and slow recovery dB/second because again, I wanted to see what was going on, and be able to demonstrate in the video?
- My video test case was also extreme... S9+30dB... to demonstrate the extreme capability of the AGC system to totally delete any possibility of that massive signal reaching the earphones, in this case it instantly applied about 36dB attenuation, and no blip at all is seen on the Audacity software's recording of the audio (which is also streamed over the USB sound card interface concurrently to the audio output). And also to clearly show everything for the video.?
- My 40m noise floor is too high here right now, as I said on the video, my antenna fell down. It doesn't work as well lying on the roof which has a metal frame inside the tiles.?
Easter egg: If you press the D key on the terminal screen, in the main menu, you get my secret "lab", it's my playground which I use during development of my various functional additions, and for debugging, etc. What's in there now is some debug info on the AGC as that's the task I had most recently on the lab bench. Might be interesting, if you are wanting to play with the AGC.?
--
Jginaz
|
On Fri, Feb 16, 2024 at 02:04 PM, <nz0tham@...> wrote:
I got 17 loaded just fine.? Anyone have suggested AGC settings to start with?
73, Bill NZ0T
See below.? The high "Smooth Samples" setting seems to eliminate much of the audio clicks.? The remaining settings give reasonably good dynamic range while limiting strong signals. I'm still experimenting with values but this is a good start. Paul, W9AC ?
|
Paul Christensen