fldigi CW decoding problems
开云体育
I'm having a problem with my fldigi (v4.2.06) and CW decoding.? During the recent CQ WW DX CW contest, I used it (connected to the club's IC-7300) and more than 80-90% of the time, even on a fairly 'strong' signal, it either wouldn't decode (gave me gibberish)
even when I had the selector (I think that's what it's called) directly ON?the signal and/or decoded the wrong call, or just part of the call, then gibberish again.? If it had decoded correctly, I
MIGHT?have been able to log quite a few more contacts than what I did (only 94 contacts total for the whole contest).
Now I realize that hand-sent CW and machine-sent CW aren't the same thing, but I'll bet that a majority of? ?CW contesters (even the DX ones) are using either fldigi or something similar to it, based on the speeds I was seeing on some of them (I don't think
there's very many CW ops that can hand send CW at more than 35 WPM, but occasionally, my fldigi was reporting 40, even 45 WPM on some potential contacts.
What fldigi setting(s) could/should I set mine for in order for it to decode correctly/properly/more accurately, without giving me much (if anything) in the way of gibberish on decodes?? In order to even
see?the signals on the waterfall, I had to set the Upper signal level (down in the lower left corner, near the "WF/FFT/SIG" button) to at least a -45 dB and when Tx'ing, my Tx level attenuator was set at (normally) -5 or -6 dB; should they be set at
those levels, or what?? Also, is there anything (settings) on the radio itself that could/should change in addition to the fldigi changes?
In the Config dialog, under Modem > CW > General, I have the following settings:
??????SOM decoding: unchecked
??????WPM Tracking: checked
??????Filter bandwidth: ~83, Matched Filt' checked
??????Rx WPM: ~67, Range is "10"
??????Transmit: ~20 char WPM
??????Default: 20, Lower Limit: 5, Upper Limit: 50
??????Use Farnsworth timing: unchecked
??????Use Wordsworth timing: unchecked
Upper right corner, Signal Tracking:
??????Attack: Med
??????Decay: Slow
So what else am I missing here?? What other thigs should/could I set differently for better (read, more accurate) decodes?
Mike, W1DGL
Prescott, AZ (DM34sm/so - "sm" is my home QTH, "so" is the club station QTH)
|
开云体育Hi Mike, I didn't take part in the contest, but I listened to a few signals. A lot of contesters would send CQ at a highish speed followed by their call at say 18 wpm. In reply you'd get the returning call at similar speed and the 599 TU at a humongous speed. I've looked at the decoding algorithm and it takes a few dots and dashes to sync up on the speed. If the speed changes like that I am surprised it did as well as it did. Phil GM3ZZA On 26 November 2024, at 02:02, "W1DGL via groups.io" <fix_broke_airplanes@...> wrote:
I'm having a problem with my fldigi (v4.2.06) and CW decoding.? During the recent CQ WW DX CW contest, I used it (connected to the club's IC-7300) and more than 80-90% of the time, even on a fairly 'strong' signal, it either wouldn't decode (gave me gibberish)
even when I had the selector (I think that's what it's called) directly ON?the signal and/or decoded the wrong call, or just part of the call, then gibberish again.? If it had decoded correctly, I
MIGHT?have been able to log quite a few more contacts than what I did (only 94 contacts total for the whole contest).
Now I realize that hand-sent CW and machine-sent CW aren't the same thing, but I'll bet that a majority of? ?CW contesters (even the DX ones) are using either fldigi or something similar to it, based on the speeds I was seeing on some of them (I don't think
there's very many CW ops that can hand send CW at more than 35 WPM, but occasionally, my fldigi was reporting 40, even 45 WPM on some potential contacts.
What fldigi setting(s) could/should I set mine for in order for it to decode correctly/properly/more accurately, without giving me much (if anything) in the way of gibberish on decodes?? In order to even
see?the signals on the waterfall, I had to set the Upper signal level (down in the lower left corner, near the "WF/FFT/SIG" button) to at least a -45 dB and when Tx'ing, my Tx level attenuator was set at (normally) -5 or -6 dB; should they be set at
those levels, or what?? Also, is there anything (settings) on the radio itself that could/should change in addition to the fldigi changes?
In the Config dialog, under Modem > CW > General, I have the following settings:
??????SOM decoding: unchecked
??????WPM Tracking: checked
??????Filter bandwidth: ~83, Matched Filt' checked
??????Rx WPM: ~67, Range is "10"
??????Transmit: ~20 char WPM
??????Default: 20, Lower Limit: 5, Upper Limit: 50
??????Use Farnsworth timing: unchecked
??????Use Wordsworth timing: unchecked
Upper right corner, Signal Tracking:
??????Attack: Med
??????Decay: Slow
So what else am I missing here?? What other thigs should/could I set differently for better (read, more accurate) decodes?
Mike, W1DGL
Prescott, AZ (DM34sm/so - "sm" is my home QTH, "so" is the club station QTH)
|
|
This is one of the side effects of "contests" and automation. Instead of folks actually sending and receiving code by hand, machines increasingly do the "work". Many hams don't realize how these side effects appear, so the are oblivious to how the machines are affecting the throughput. It's one of the reasons I don't contest. I am not against contesting, per se. YMMV, TEHO, and all that. I appreciate you taking the time to listen to the signals, and report what you hear. The comment about software not changing gears as fast as the human brain is especially important.
toggle quoted message
Show quoted text
73 Rich NE1EE The Dusty Key On the banks of the Piscataqua On 2024-11-26 03:32:, Philip Rose via groups.io wrote:
I didn't take part in the contest, but I listened to a few signals. |
|
Hi there Mike
?
?some thoughts based on my self learning, on how to use FL-DIGI with N1MM for contesting.
1- Some signals come in so strong and distorted ( linear amplifiers not properly set) that the softwware cannot identify dots and dashes properly, it comes all as a buch of misunterprataed data. For these one, you either get them by ear or via the assitent (Cluster). The same is valid for bad keyed caractheres ( no proper space between the dash/dot, etc....) these ones you will have to use your ear.
2- Now there are some settings will help you to "clarify" signals specially when there are multiple stations very close one to each other. I use to run the MODEM setting with the filter bandwith slider always open so you can change filter on the fly? to narrow or enlarge reception.
give it a try for "SOM decoding"? setting it provides a fuzzy logic implementation to match the RX stream detected on-off sequence to a "best fit" character. It can increase the probability of correctly identifying the text character under very noisy conditions.
You can find a lot of info on settings here you will need to play around to fins the best set for your rig.
3- Finally people use different tools for contesting N1MM plus FLDigi seems the most common, some people use SDR TX with CW Skimmer, some use DM-780, some decode by ear at 50wpm .
?
Hope it help you somehow!
?
GL es HPE CU on the air, best 73.
?
Enjoy CW and Contesting.
?
De? F4WEM / PY3RK |
|
A different question
?
My FLDIGI (rig is Icom 7610) copies CW almost 100% from 10wpm to 30 wpm, if sent with appropriate timing.
Hand sent CW on a straight key is awful
?
The only thing that I can misadjust to not copy is to “enable” the FLDIGI digital squelch and set the slider very high, in the white area.
?
So be sure that the Fldigi digital squelch is turned OFF or disabled (light not on).
?
My FLDIGI copies CW when the radio’s mode is in CW, or CW reversed, or even in USB mode as long as the pitch is adjusted to where I like it - around 800 Hz.
?
All of the other CW adjustments in the software are just fine-tuning.
?
de k3eui Barry |
|
Mike, Dittos on what Barry said. I have a few other things to add from experience. Can't emphasize enough on the squelch. It can be a killer. At the bottom of the screen, Make sure your CW speed is set to plus or minus what you may think the speed is. If they are kicking 35 wpm and you have it set at 10 or 15, you'll never decode it. Just to the left of the squelch, you have your audio adjustment. You want to make sure the audio coming into FLDIGI isn't distorted and that adjustment there to the left of the squelch will help you with that. 73 es GL John KC4LZN On Thu, Dec 5, 2024 at 7:33?AM K3EUI Barry via <k3euibarry=[email protected]> wrote:
|
|
A study of the on-line documentation for starters. Forget trying to decode hand sent elements. Computer sent elements decode just fine. There is no anticipatory algorithm that tells FLDIGI what's coming next. Head copy is always superior to computer copy. If you are expecting the computer to do the decoding for all CW signals, forget it, it won't happen. Good code equals good copy. Poor sent code equal's poor copy.
Don't get the receiver bandwidth too narrow as filters can ring, thus contribute to poor copy. I normally run 1.0 to 1.5 kHz receiver BW. I find FLDIGI to be one of the better CW decoding applications. In fact I use it with N1MM as the CW engine. I have HRD, MixW4, CW Get, and a couple of other CW applications. Each has their merits. My Elecraft K3S does an excellent job decoding CW signals........if the parameters, threshold, etc. are correctly adjusted. It takes patience and lots of trial and error, along with understanding of signal and band noise relationships. There is NO one setting that will make it always great. 73 Bob, K4TAX |
开云体育
Thanks Bob,
Once upon a time I tried to get Fldigi to copy my hand sent morse. I eventually managed it. I've looked at the algorithm and indeed have coded into a keyer/decoder I built using an Arduino. The algorithm works fine with a pure digital on-off signal. However
trying to isolate the on-off signal from the audio pass-band was more problematic. The DSP requirements were beyond the capabilities of the Arduino.
Your comment on the effects of the filter are very relevant. When I use fldigi on-air as a back-up to head copy, it fails dramatically. Head copy requires a narrowish filter. Too narrow and the ringing even makes head copy difficult. Fldigi, as you said, needs
a wider filter.
73 Phil GM3ZZA
From: [email protected] <[email protected]> on behalf of Bob McGraw - K4TAX <rmcgraw@...>
Sent: 24 January 2025 7:13 PM To: [email protected] <[email protected]> Subject: Re: [winfldigi] fldigi CW decoding problems ?
A study of the on-line documentation for starters.?? Forget trying to decode hand sent elements.? Computer sent elements decode just fine.?? There is no anticipatory algorithm that tells FLDIGI what's coming next.?? Head copy is always
superior to computer copy.?? If you are expecting the computer to do the decoding for all CW signals, forget it, it won't happen.?? Good code equals good copy.? Poor sent code equal's poor copy.
Don't get the receiver bandwidth too narrow as filters can ring, thus contribute to poor copy.? I normally run 1.0 to 1.5 kHz receiver BW. I find FLDIGI to be one of the better CW decoding applications.? In fact I use it with N1MM as the CW engine.?? I have HRD, MixW4, CW Get, and a couple of other CW applications.?? Each has their merits.???? My Elecraft K3S does an excellent job decoding CW signals........if the parameters, threshold, etc. are correctly adjusted.? It takes patience and lots of trial and error, along with understanding of signal and band noise relationships.?? There is NO one setting that will make it always great. 73 Bob, K4TAX |
|
RE: No decode on hand sent CW from the K4. Is the monitor audio getting to the computer? I just checked my K3S feeding my computer as normal and FLDIGI does decode my keyer sent CW. It also decodes my hand sent CW with my straight key. Check your monitor level audio going to the computer and application.
73 Bob, K4TAX |
|
I use PCW Fistcheck all the time to keep in practice twixt on air sessions. I find that it does really well at decoding what my rigs send. I use a straight key 90% of the time, which is why I like the PCWF feedback on my timing. The timing dialog is great for that. I agree that getting CW off the air is another matter, with signal shifts, fading, etc, but I have used PCFW with a local ham on HF, and it was pretty good. In fact, PCWF showed what fldigi could not...
toggle quoted message
Show quoted text
One of the new CW hams was using a keyer, and thinking that he was sending great code. His dits and dahs were great, actually...as you'd expect from a keyer. Problem was that he couldn't send a letter or a word that was generally understandable by ear, so PCWF had no chance. But I copied what PCWF recorded, and sent that to him. He got the message right away, and decided to shift from keyer to straight key until he had a better sense of timing. I think using fldigi is overkill for these simple tasks, but I understand wanting to use it if the goal is to actually send and receive CW using fldigi. I occasionally use Morse Expert. It is great at showing just how much the RX CW signal shifts around, and why it is so difficult to machine decode hand sent CW. 73 Rich NE1EE The Dusty Key On the banks of the Piscataqua On 2025-01-25 10:54:, Bob McGraw - K4TAX wrote:
RE: No decode on hand sent CW from the K4. Is the monitor audio getting to the computer? |
|
Check the Squelch setting, being sure the signal is above the squelch bar. Or just turn off squelch. Also make sure the RX speed is equal to or greater than the speed being sent.
On the other hand, if your keying is good then the application will decode correctly. If it doesn't decode correctly, well,...... perhaps your keying needs improvement. 73 Bob, K4TAX |
W1DGL
Philip Rose
Rich NE1EE
PY3RK / F4WEM
K3EUI Barry
John KC4LZN
Bob McGraw - K4TAX