I have been doing some evaluations of Linsim and I have identified a couple of minor issues.?
In my early tests, I selected 1 path and a spread of 1Hz and and AWGN with SNR=25dB.?
I did one set of tests at a sample rate of 8kbs and another set at 48kbs. I repeated the tests?
with path 1 selected, a spread of 1Hz, and an offset of +5Hz.?

?

I am using Windows 11+Dell G10 laptop.

?

Observations

?

1. I intercomapared Linsim v2.0.4 and and Moe Wheatley's Pathsim v1.0. I generated a 1kHz sinewave
with a 60 sec duration at a sample rate of 8kbs. In this case the time domain wave forms exhibited
very similar envelope fluctuations. When I performed an FFT on the records the resulting spectra
exhibited a Gaussian shape with the same width. When I applied an offset of +5.0Hz, the spectrum
shifted up by 5.0Hz with Pathsim. With Linsim, the Gaussian spectrum shifted down by -5.0 Hz. So there?
is a potential sign error with the offset in the Linsim code.

?

2. I performed an additional set of tests with a 1kHz sinewave with a 60sec duration and a sample
rate of 48kbs. In this case I did not use Pathsim since it only works with a sample rate of 8kbs.
I first applied the 1kHz sinewave to Linsim with Path 1 selected, a spread of 1 Hz, and 0 offset.
The resulting output waveform exhibited much faster envelope fluctuations. This was confirmed?
with the FFT, which showed markedly wider Gaussian spectrum-the width increased approximately 6x.

?

3. I suspected that this spread issue has something to do with the different data rates since is 6x
larger than the 8kbs case. I then repeated the test with a spread of 1/6 = 0.166, that?
corresponds to the ratio of the data rates 8kbs/48kbs. The resulting spectrum using an FFT showed?
the correct shape and frequency spread.

?

4. I think this spreading issue can be corrected in Linsim by applying a scale factor to the
?spread that accounts for the ratio: 8kbs/(the selected date rate kbs)--My best guess :-)
?
5. When I applied a +5.0 Hz offset with the 48kbs sinewave the resulting spectrum shifted down
-5.0 Hz. So this Offset sign error is the same at the higher sample rates.?

?

Summary

?

1. Linsim is a robust and much needed update to Pathsim that enables processing of .wav files?
for a variety of data rates in the range 8kbs to 48 kbs.

?

2. When using Linsim at data rates other than 8kbs, adjust the spread using a scale factor of
8kbs/(the selected data rate kbs).

?

3. There is a sign error in the offset. So simply switch the sign of the desired frequency offset.

?

4. I have applied pulse trains generated with Matlab to Linsim and I will report on my findings?
soon.

?

Bob Johnk, WB0VGD?

Arvada, Colorado

I have been doing some evaluations of Linsim and I have identified a couple of minor issues.?
In my early tests, I selected 1 path and a spread of 1Hz and and AWGN with SNR=25dB.?
I did one set of tests at a sample rate of 8kbs and another set at 48kbs. I repeated the tests?
with path 1 selected, a spread of 1Hz, and an offset of +5Hz.?

?

I am using Windows 11+Dell G10 laptop.

?

Observations

?

1. I intercomapared Linsim v2.0.4 and and Moe Wheatley's Pathsim v1.0. I generated a 1kHz sinewave
with a 60 sec duration at a sample rate of 8kbs. In this case the time domain wave forms exhibited
very similar envelope fluctuations. When I performed an FFT on the records the resulting spectra
exhibited a Gaussian shape with the same width. When I applied an offset of +5.0Hz, the spectrum
shifted up by 5.0Hz with Pathsim. With Linsim, the Gaussian spectrum shifted down by -5.0 Hz. So there?
is a potential sign error with the offset in the Linsim code.

?

2. I performed an additional set of tests with a 1kHz sinewave with a 60sec duration and a sample
rate of 48kbs. In this case I did not use Pathsim since it only works with a sample rate of 8kbs.
I first applied the 1kHz sinewave to Linsim with Path 1 selected, a spread of 1 Hz, and 0 offset.
The resulting output waveform exhibited much faster envelope fluctuations. This was confirmed?
with the FFT, which showed markedly wider Gaussian spectrum-the width increased approximately 6x.

?

3. I suspected that this spread issue has something to do with the different data rates since is 6x
larger than the 8kbs case. I then repeated the test with a spread of 1/6 = 0.166, that?
corresponds to the ratio of the data rates 8kbs/48kbs. The resulting spectrum using an FFT showed?
the correct shape and frequency spread.

?

4. I think this spreading issue can be corrected in Linsim by applying a scale factor to the
?spread that accounts for the ratio: 8kbs/(the selected date rate kbs)--My best guess :-)
?
5. When I applied a +5.0 Hz offset with the 48kbs sinewave the resulting spectrum shifted down
-5.0 Hz. So this Offset sign error is the same at the higher sample rates.?

?

Summary

?

1. Linsim is a robust and much needed update to Pathsim that enables processing of .wav files?
for a variety of data rates in the range 8kbs to 48 kbs.

?

2. When using Linsim at data rates other than 8kbs, adjust the spread using a scale factor of
8kbs/(the selected data rate kbs).

?

3. There is a sign error in the offset. So simply switch the sign of the desired frequency offset.

?

4. I have applied pulse trains generated with Matlab to Linsim and I will report on my findings?
soon.

?

Bob Johnk, WB0VGD?

Arvada, Colorado

John,

The computational part of linsim is modeled exactly after pathsim.? All of the path, noise etc is performed at 8000 samples per second data.? Source and output file sample rates are converted using high precision resampling libraries.? I want to duplicate your tests.? For the 48kbs tests did you use both input and output files to 48kbs?? linsim can be used as a rate converter if no other signal processing is performed.

David

On 1/11/25 12:58, Robert Johnk via groups.io wrote:

I have been doing some evaluations of Linsim and I have identified a couple of minor issues.?
In my early tests, I selected 1 path and a spread of 1Hz and and AWGN with SNR=25dB.?
I did one set of tests at a sample rate of 8kbs and another set at 48kbs. I repeated the tests?
with path 1 selected, a spread of 1Hz, and an offset of +5Hz.?

?

I am using Windows 11+Dell G10 laptop.

?

Observations

?

1. I intercomapared Linsim v2.0.4 and and Moe Wheatley's Pathsim v1.0. I generated a 1kHz sinewave
with a 60 sec duration at a sample rate of 8kbs. In this case the time domain wave forms exhibited
very similar envelope fluctuations. When I performed an FFT on the records the resulting spectra
exhibited a Gaussian shape with the same width. When I applied an offset of +5.0Hz, the spectrum
shifted up by 5.0Hz with Pathsim. With Linsim, the Gaussian spectrum shifted down by -5.0 Hz. So there?
is a potential sign error with the offset in the Linsim code.

?

2. I performed an additional set of tests with a 1kHz sinewave with a 60sec duration and a sample
rate of 48kbs. In this case I did not use Pathsim since it only works with a sample rate of 8kbs.
I first applied the 1kHz sinewave to Linsim with Path 1 selected, a spread of 1 Hz, and 0 offset.
The resulting output waveform exhibited much faster envelope fluctuations. This was confirmed?
with the FFT, which showed markedly wider Gaussian spectrum-the width increased approximately 6x.

?

3. I suspected that this spread issue has something to do with the different data rates since is 6x
larger than the 8kbs case. I then repeated the test with a spread of 1/6 = 0.166, that?
corresponds to the ratio of the data rates 8kbs/48kbs. The resulting spectrum using an FFT showed?
the correct shape and frequency spread.

?

4. I think this spreading issue can be corrected in Linsim by applying a scale factor to the
?spread that accounts for the ratio: 8kbs/(the selected date rate kbs)--My best guess :-)
?
5. When I applied a +5.0 Hz offset with the 48kbs sinewave the resulting spectrum shifted down
-5.0 Hz. So this Offset sign error is the same at the higher sample rates.?

?

Summary

?

1. Linsim is a robust and much needed update to Pathsim that enables processing of .wav files?
for a variety of data rates in the range 8kbs to 48 kbs.

?

2. When using Linsim at data rates other than 8kbs, adjust the spread using a scale factor of
8kbs/(the selected data rate kbs).

?

3. There is a sign error in the offset. So simply switch the sign of the desired frequency offset.

?

4. I have applied pulse trains generated with Matlab to Linsim and I will report on my findings?
soon.

?

Bob Johnk, WB0VGD?

Arvada, Colorado

I have been doing some evaluations of Linsim and I have identified a couple of minor issues.?
In my early tests, I selected 1 path and a spread of 1Hz and and AWGN with SNR=25dB.?
I did one set of tests at a sample rate of 8kbs and another set at 48kbs. I repeated the tests?
with path 1 selected, a spread of 1Hz, and an offset of +5Hz.?

?

I am using Windows 11+Dell G10 laptop.

?

Observations

?

1. I intercomapared Linsim v2.0.4 and and Moe Wheatley's Pathsim v1.0. I generated a 1kHz sinewave
with a 60 sec duration at a sample rate of 8kbs. In this case the time domain wave forms exhibited
very similar envelope fluctuations. When I performed an FFT on the records the resulting spectra
exhibited a Gaussian shape with the same width. When I applied an offset of +5.0Hz, the spectrum
shifted up by 5.0Hz with Pathsim. With Linsim, the Gaussian spectrum shifted down by -5.0 Hz. So there?
is a potential sign error with the offset in the Linsim code.

?

2. I performed an additional set of tests with a 1kHz sinewave with a 60sec duration and a sample
rate of 48kbs. In this case I did not use Pathsim since it only works with a sample rate of 8kbs.
I first applied the 1kHz sinewave to Linsim with Path 1 selected, a spread of 1 Hz, and 0 offset.
The resulting output waveform exhibited much faster envelope fluctuations. This was confirmed?
with the FFT, which showed markedly wider Gaussian spectrum-the width increased approximately 6x.

?

3. I suspected that this spread issue has something to do with the different data rates since is 6x
larger than the 8kbs case. I then repeated the test with a spread of 1/6 = 0.166, that?
corresponds to the ratio of the data rates 8kbs/48kbs. The resulting spectrum using an FFT showed?
the correct shape and frequency spread.

?

4. I think this spreading issue can be corrected in Linsim by applying a scale factor to the
?spread that accounts for the ratio: 8kbs/(the selected date rate kbs)--My best guess :-)
?
5. When I applied a +5.0 Hz offset with the 48kbs sinewave the resulting spectrum shifted down
-5.0 Hz. So this Offset sign error is the same at the higher sample rates.?

?

Summary

?

1. Linsim is a robust and much needed update to Pathsim that enables processing of .wav files?
for a variety of data rates in the range 8kbs to 48 kbs.

?

2. When using Linsim at data rates other than 8kbs, adjust the spread using a scale factor of
8kbs/(the selected data rate kbs).

?

3. There is a sign error in the offset. So simply switch the sign of the desired frequency offset.

?

4. I have applied pulse trains generated with Matlab to Linsim and I will report on my findings?
soon.

?

Bob Johnk, WB0VGD?

Arvada, Colorado