[digitalradio] Re: S/N Multipsk figures -- JT65A vs Olivia and others
I'm perplexed by the edited dB figures. On JT65A HF it doesn't make any sense that the values are -5 or -6 db when the signal is strong and moving the S-meter to s5 or s6. Here we use a 300Hz filter and the audio output is adjusted to read about 0db with no signal. What I would have expected is the db value would be referenced to this receiver noise floor value. Secondly, I can hear and copy the CW ID at edited values of -20db or so. That also makes no sense. This threshold should be around -10db or so below the RX noise floor. This audible threshold is pretty much independent of whether one uses a 2.1 KHz filter or the narrower 300HZ filter. In fact, if I'm interested in hearing really weak signals, using the 2.1KHz filter allows weaker CW signals to be heard -- presumably due to less attenuation in the wider filter. This only works of course if there is no signal within the filter passband which starts AGC action. AVC use no doubt confounds things for stronger signals. However, there is no option on the IC706 to turn of AGC. Audio output is pretty much linear (as per ARRL BPL studies) below the AGC threshold. I estimate that threshould to be about S2 for the 706. So just what does the edited db value mean? It certainly does not represent how far the signal is below the RX noise floor. Since RX gain is unknown, it can't represent some absolute value of voltage. 73 de Brian/K3KO --- In digitalradio@yahoogroups.com, Patrick Lindecker [EMAIL PROTECTED] wrote: Hello to all, Comparizon with S/N Multipsk figures. According to JT65 specifications, this mode decodes with few errors down to -23 dB, with a normalized band of 2.5 KHz. All Multipsk figures are normalized with a band of 3 KHz. -23 dB in 2.5 KHz is about -24 dB in 3 KHz band (-23.792 dB exactly). This figure of -24 dB can be compared to Olivia 250-8 which has a minimum S/N of -14 dB. So JT65 is 10 dB better or 10 times better. But of course JT65 is much slower that Olivia 250-8. The only modes which are close to JT65 are: * THROBX: Lowest S/N: -18,5 dB for the 1 baud, -17.5 dB for the 2 bauds * PSKAM10: Lowest S/N : -19.5 dB In conclusion JT65 is better (under S/N criteria) that any modes in Multipsk. 73 Patrick - Original Message - From: Tony To: digitalradio@yahoogroups.com Sent: Saturday, April 21, 2007 9:25 AM Subject: [digitalradio] Path Simulator tests -- JT65A vs Olivia and others All: I used Pathsim to compare the sensitivity of JT65A vs MFSK, PSK31 and OLIVIA using AWGN to alter the SNR. I ran direct-path with no ionospheric disturbance. The chat modes decoded with error-free print down to -12 to -14db SNR. The JT65A mode decoded at -27db SNR (signal inaudible). Assuming the Pathsim white noise measurments were accurate, I think it's safte to say that JT65 is capable of decoding much weaker signals than the others. Would be interesting to see how it does with simulated ionospheric disturbances. 73 Tony - KT2Q
Re: [digitalradio] Re: S/N Multipsk figures -- JT65A vs Olivia and others
Hello brian, The S/N referenced to a bandwidth is used to compare modes under a noise environment criteria. A minimum S/N of 0 dB means that with an equal power (let's say 1 watt) of signal and noise (noise distributed over a 3 KHz band so with a density of 0.33 W/KHz), the signal transmitted will be decoded. In an other mode with a minimum S/N of -10 dB, you will need only 0.1 Watt of signal for 1 Watt of the same noise to decode the text transmitted, and so on. In fact, if I'm interested in hearing really weak signals, using the 2.1KHz filter allows weaker CW signals to be heard -- The ability to decode a weak (CW) signal is a psycho-acoustic problem. As far as I know, reducing the bandwidth helps down to a limit (perhaps 500 Hz?). 73 Patrick - Original Message - From: Brian A To: digitalradio@yahoogroups.com Sent: Saturday, April 21, 2007 2:07 PM Subject: [digitalradio] Re: S/N Multipsk figures -- JT65A vs Olivia and others I'm perplexed by the edited dB figures. On JT65A HF it doesn't make any sense that the values are -5 or -6 db when the signal is strong and moving the S-meter to s5 or s6. Here we use a 300Hz filter and the audio output is adjusted to read about 0db with no signal. What I would have expected is the db value would be referenced to this receiver noise floor value. Secondly, I can hear and copy the CW ID at edited values of -20db or so. That also makes no sense. This threshold should be around -10db or so below the RX noise floor. This audible threshold is pretty much independent of whether one uses a 2.1 KHz filter or the narrower 300HZ filter. In fact, if I'm interested in hearing really weak signals, using the 2.1KHz filter allows weaker CW signals to be heard -- presumably due to less attenuation in the wider filter. This only works of course if there is no signal within the filter passband which starts AGC action. AVC use no doubt confounds things for stronger signals. However, there is no option on the IC706 to turn of AGC. Audio output is pretty much linear (as per ARRL BPL studies) below the AGC threshold. I estimate that threshould to be about S2 for the 706. So just what does the edited db value mean? It certainly does not represent how far the signal is below the RX noise floor. Since RX gain is unknown, it can't represent some absolute value of voltage. 73 de Brian/K3KO --- In digitalradio@yahoogroups.com, Patrick Lindecker [EMAIL PROTECTED] wrote: Hello to all, Comparizon with S/N Multipsk figures. According to JT65 specifications, this mode decodes with few errors down to -23 dB, with a normalized band of 2.5 KHz. All Multipsk figures are normalized with a band of 3 KHz. -23 dB in 2.5 KHz is about -24 dB in 3 KHz band (-23.792 dB exactly). This figure of -24 dB can be compared to Olivia 250-8 which has a minimum S/N of -14 dB. So JT65 is 10 dB better or 10 times better. But of course JT65 is much slower that Olivia 250-8. The only modes which are close to JT65 are: * THROBX: Lowest S/N: -18,5 dB for the 1 baud, -17.5 dB for the 2 bauds * PSKAM10: Lowest S/N : -19.5 dB In conclusion JT65 is better (under S/N criteria) that any modes in Multipsk. 73 Patrick - Original Message - From: Tony To: digitalradio@yahoogroups.com Sent: Saturday, April 21, 2007 9:25 AM Subject: [digitalradio] Path Simulator tests -- JT65A vs Olivia and others All: I used Pathsim to compare the sensitivity of JT65A vs MFSK, PSK31 and OLIVIA using AWGN to alter the SNR. I ran direct-path with no ionospheric disturbance. The chat modes decoded with error-free print down to -12 to -14db SNR. The JT65A mode decoded at -27db SNR (signal inaudible). Assuming the Pathsim white noise measurments were accurate, I think it's safte to say that JT65 is capable of decoding much weaker signals than the others. Would be interesting to see how it does with simulated ionospheric disturbances. 73 Tony - KT2Q
[digitalradio] Re: S/N Multipsk figures -- JT65A vs Olivia and others
Thank you Patrick for the explanation. No doubt the mode has solid theoretical resons for what it edits. Intercomparing various digital modes has some merit-- like for a Ph.D thesis or marketing. However as a user, the present value edited isn't a number that is too useful. It apparently tells me nothing about how far below the RX noise the signal I'm copying really is. That's unfortunate. Making contacts with large -db values seems to overstate the mode's capabilities in this regard. It's kind of like the expensive receivers which show S9 signal levels for weak stations-- making the purchaser happy he spend the extra money. Then later he finds out that the manufacturer has calibrated his S meter in 3db (or less) increments and his receiver is perhaps only marginally better. For me these 6 + minute/QSO's are agony. I'm really after some indication of what the agony is really buying in signal reception. 73 de Brian/K3KO --- In digitalradio@yahoogroups.com, Patrick Lindecker [EMAIL PROTECTED] wrote: Hello brian, The S/N referenced to a bandwidth is used to compare modes under a noise environment criteria. A minimum S/N of 0 dB means that with an equal power (let's say 1 watt) of signal and noise (noise distributed over a 3 KHz band so with a density of 0.33 W/KHz), the signal transmitted will be decoded. In an other mode with a minimum S/N of -10 dB, you will need only 0.1 Watt of signal for 1 Watt of the same noise to decode the text transmitted, and so on. In fact, if I'm interested in hearing really weak signals, using the 2.1KHz filter allows weaker CW signals to be heard -- The ability to decode a weak (CW) signal is a psycho-acoustic problem. As far as I know, reducing the bandwidth helps down to a limit (perhaps 500 Hz?). 73 Patrick - Original Message - From: Brian A To: digitalradio@yahoogroups.com Sent: Saturday, April 21, 2007 2:07 PM Subject: [digitalradio] Re: S/N Multipsk figures -- JT65A vs Olivia and others I'm perplexed by the edited dB figures. On JT65A HF it doesn't make any sense that the values are -5 or -6 db when the signal is strong and moving the S-meter to s5 or s6. Here we use a 300Hz filter and the audio output is adjusted to read about 0db with no signal. What I would have expected is the db value would be referenced to this receiver noise floor value. Secondly, I can hear and copy the CW ID at edited values of -20db or so. That also makes no sense. This threshold should be around -10db or so below the RX noise floor. This audible threshold is pretty much independent of whether one uses a 2.1 KHz filter or the narrower 300HZ filter. In fact, if I'm interested in hearing really weak signals, using the 2.1KHz filter allows weaker CW signals to be heard -- presumably due to less attenuation in the wider filter. This only works of course if there is no signal within the filter passband which starts AGC action. AVC use no doubt confounds things for stronger signals. However, there is no option on the IC706 to turn of AGC. Audio output is pretty much linear (as per ARRL BPL studies) below the AGC threshold. I estimate that threshould to be about S2 for the 706. So just what does the edited db value mean? It certainly does not represent how far the signal is below the RX noise floor. Since RX gain is unknown, it can't represent some absolute value of voltage. 73 de Brian/K3KO --- In digitalradio@yahoogroups.com, Patrick Lindecker f6cte@ wrote: Hello to all, Comparizon with S/N Multipsk figures. According to JT65 specifications, this mode decodes with few errors down to -23 dB, with a normalized band of 2.5 KHz. All Multipsk figures are normalized with a band of 3 KHz. -23 dB in 2.5 KHz is about -24 dB in 3 KHz band (-23.792 dB exactly). This figure of -24 dB can be compared to Olivia 250-8 which has a minimum S/N of -14 dB. So JT65 is 10 dB better or 10 times better. But of course JT65 is much slower that Olivia 250-8. The only modes which are close to JT65 are: * THROBX: Lowest S/N: -18,5 dB for the 1 baud, -17.5 dB for the 2 bauds * PSKAM10: Lowest S/N : -19.5 dB In conclusion JT65 is better (under S/N criteria) that any modes in Multipsk. 73 Patrick - Original Message - From: Tony To: digitalradio@yahoogroups.com Sent: Saturday, April 21, 2007 9:25 AM Subject: [digitalradio] Path Simulator tests -- JT65A vs Olivia and others All: I used Pathsim to compare the sensitivity of JT65A vs MFSK, PSK31 and OLIVIA using AWGN to alter the SNR. I ran direct-path with no ionospheric disturbance. The chat modes decoded with error-free print down to -12 to -14db SNR. The JT65A mode decoded at -27db SNR (signal inaudible