random letters
david clifford
Dick Wilkins
Hello from a cool n windy North Wales. i have got my psk 31 to tx and rx i am using airlink express and i have made a couple of contacts. i would like to fine tune the system, can any tell me why i get so many random letters on my screen and what can i do to minimise this. many thanks. David C G0NVF
--
--
http://bpsk31.com
You received this message because you are a member of the bpsk31.com discussion list.
To post to this group, send email to bps...@googlegroups.com
To unsubscribe from this group, send email to bpsk31+unsubscribe@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/bpsk31?hl=en?hl=en
---
You received this message because you are subscribed to the Google Groups "bpsk31.com Discussion list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bpsk31+unsubscribe@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
jimfraziernm
--
http://bpsk31.com
You received this message because you are a member of the bpsk31.com discussion list.
To post to this group, send email to bps...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/bpsk31?hl=en?hl=en
---
You received this message because you are subscribed to the Google Groups "bpsk31.com Discussion list" group.
Gary Peach
VERY interesting, ( and very good advice), please may I offer this TORCH to illuminate this discussion and to be of more GENERAL Interest
Gary, G7SLL
PS NB to work 100% error free the signal to noise ratio must be greater than 3.5 dB barely [from the table} 0.01 [S+N]/N on your S-Meter
Note The actual effective bandwidth in PSK31 is about one Hz, 😊 thus noise is kTB is effectively kT where T is the SKY temperature
This term kT is called the “noise spectral density”
You want to work a spacecraft out at Jupiter signal level less that – 174 dBm (or – S17 minus below the zero on your S METER) with data better than one error in a million ( similar decoding but comparison to a dictionary to verify the veracity of the data.
That’s a whole new area of interest and something for the aficionados to start work on developing for HAM use. 😊
START HERE
Errors correlated to Signal to Noise Ratio
|
S-METER |
actual |
BLOCK size |
|||||
|
[S+N]/N |
S/N |
errors |
10 CHR |
20 CHR |
100 CHR |
1,000 CHR |
|
|
0.63 dB |
0.0 dB |
1.000000 |
15.73% |
2 err |
3 err |
16 err |
157 err |
|
0.46 dB |
0.5 dB |
1.122018 |
11.26% |
1 err |
2 err |
11 err |
113 err |
|
0.31 dB |
1.0 dB |
1.258925 |
7.50% |
1 err |
2 err |
8 err |
75 err |
|
0.19 dB |
1.5 dB |
1.412538 |
4.58% |
0 err |
1 err |
5 err |
46 err |
|
0.11 dB |
2.0 dB |
1.584893 |
2.50% |
0 err |
1 err |
3 err |
25 err |
|
0.05 dB |
2.5 dB |
1.778279 |
1.19% |
0 err |
0 err |
1 err |
12 err |
|
0.02 dB |
3.0 dB |
1.995262 |
0.48% |
0 err |
0 err |
0 err |
5 err |
|
0.01 dB |
3.5 dB |
2.238721 |
0.15% |
0 err |
0 err |
0 err |
2 err |
|
0.00 dB |
4.0 dB |
2.511886 |
0.04% |
0 err |
0 err |
0 err |
0 err |
|
0.00 dB |
4.5 dB |
2.818383 |
0.01% |
0 err |
0 err |
0 err |
0 err |
|
0.00 dB |
5.0 dB |
3.162278 |
0.00% |
0 err |
0 err |
0 err |
0 err |
What you see on your S METER is Signal PLUS NOISE to Noise Ratio. Left column [S+N]/N
Our interest is what the detector decoder has to deal with signal to noise. at any moment is what is being presented true or false. Counting good and bad we form two columns and can calculate the ratio between good and bad to produce the SIGNAL to NOISE RATIO. S/N
Long hand (S/N) + (N/N) = (S/N) + 1
All is revealed we can now extract the Ratio Signal to Noise simply ~~~ ([S+N]/N) -1 = S/N the thing that we want to know.
NOISE is always present; the only factor is how much noise is there. Sitting quietly at night in a country cottage, a fireside conversation is only punctuated by the crackle of burning timber in the hearth. Seated each side of the fire separated some ten feet distant conversation is easy at fairly low level.
Try to replicate that in a city flat adjacent to a busy main road and two things will happen the windows will be closed, the distance between participants will decrease. Speech will be at a higher volume closer to shouting. Closer attention is required to catch most of the conversation, even so despite paying more exhausting attention some bits will be lost of misunderstood.
Now take the conversation to the day, out into the high street. It will take place at a slower rate, the complexity of the intelligence to be passed will reduce. Even sweet nothings whispered directly into the ear will be difficult to copy.
Communication in the council chamber is limited by the barracking of the opposition and making a political point becomes more difficult. Such is democracy.
A speaker reads from his notes and the audience attempt to write down what the Speaker has said. Comparing the original with the various recorded versions we have several tables of errors. Varuopus comparisons my be made how far the listener was from the speaker. Was the opposition positioned such that the listener heard more of the heckling. A whole host of variables. That is noise.
The this data when examined on many occasions fist into the bell shaped normal distribution curve. This phenomena exist for everything that you do in life. A simple example put two pencil dots on a piece of plain paper. Using a simple school ruler measure and record the direct distance between the dots ten times and without cheat record the ten values.
It has been found by the repetition of this type of exercise many time that the data fits the Normal distribution bell shaped curve.
To illustrate ( draws curve,} limited by simple plain text formats used in groups.
However, tables have been previously calculated and recorded. This are available in various forms at the stationers.
https://www.geophysik.uni-muenchen.de/~malservisi/GlobaleGeophysik2/erf_tables.pdf
we have the tools we will rebuild our proposition
a piece of text transmitted in a noisy environment will be reproduced at the receiver with a number of error depending upon the signal to noise ratio of the received signal.
Expressed mathematically ~~~~ number of errors = 1- erf(S/N)
Erf is the value that you look up in your normal distribution table
For those with Microsoft Excel this function is built in and may be used in the same way that you would use LOG, SIN, COS etc.
However, exercise caution that you do use signal to noise ratio and NOT [S+N]/N
Having gone this far I’ll finish this off with the fact that using this information and any of the standard decoding programs. Ifa standard block of characters is transmitted and the number of errors counted then by applying the theory in revers we may measure the signal to noise ratio. Give a sufficient large block of data then the precision with which this measurement may be made may have any desirable precision that you want.
For propagation studies this enables precision to noise and NOPT the value directly established from the S Meter.
END for now
Gary Peach
The actual signal is well below Squelch Level you’ll loose a lot of possible contacts.
If possible turn OFF the AGC, and perhaps apply the ATTENUATOR but be gentle don’t strangle the poor thing
This is where an attenuator with finer than 10 dB steps would help.
Turning off the AGC removes
the 1-H(s) S/N reduction, PROBLEM.
GENERAL Communications Receivers are built to cover the very old fashioned anticipated possible signals. Therefore they may not perform ideally for any specific application. The industry has blithely carried on with philosophies laid down since the very beginning. Now that we have the means the entire basis for receiver design is up for review.
e. g.
ALL the modern RX have far too much gain at the low frequencies . that could be addressed in a much better way than by just adding an attenuator.
Old theory says it doesn’t matter if the bandwidth of the system is applied at the front or the Back. A little thought will reveal that, that old idea is a very bad blunder.
All RX would benefit from a three stage (minimum 18 dB/Octave roll off, (More would be better) { steep sided filters give best S/N} ), band pass filter at signal frequency between AE and RX. Most front ends are actually a couple of MHz wide and the noise swamps the emitter base somewhere about mid IF amplifier.
From: bps...@googlegroups.com [mailto:bps...@googlegroups.com]
On Behalf Of Dick Wilkins
Sent: 12 June 2017 20:05
To: bps...@googlegroups.com
Subject: Re: [bpsk31] random letters
Hi I would try turning the squelch up to cut off the noise when listening. KC6HBB Sam. Hope this helps.
On Jun 12, 2017 1:02 PM, "david clifford" <davy...@gmail.com> wrote:
Hello from a cool n windy North Wales. i have got my psk 31 to tx and rx i am using airlink express and i have made a couple of contacts. i would like to fine tune the system, can any tell me why i get so many random letters on my screen and what can i do to minimise this. many thanks. David C G0NVF
--
--
http://bpsk31.com
You received this message because you are a member of the bpsk31.com discussion list.
To post to this group, send email to bps...@googlegroups.com
To unsubscribe from this group, send email to bpsk31+un...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/bpsk31?hl=en?hl=en
---
You received this message because you are subscribed to the Google Groups "bpsk31.com Discussion list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bpsk31+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
--
--
http://bpsk31.com
You received this message because you are a member of the bpsk31.com discussion list.
To post to this group, send email to bps...@googlegroups.com
To unsubscribe from this group, send email to bpsk31+un...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/bpsk31?hl=en?hl=en
---
You received this message because you are subscribed to the Google Groups "bpsk31.com Discussion list" group.
To unsubscribe from this group and stop receiving emails from it, send an email to bpsk31+un...@googlegroups.com.