linrad support: Calibrate with a weak pulse.
(Sept 30 2007)

Hardware and raw data.

The calibration pulses are contained in http://www.sm5bsz.com/linuxdsp/flat/cal3/blankerinit3.raw (5 424 761 bytes) This file was recorded with the WSE converters and a Delta 44 soundcard. The calibration pulses is the square wave TTL output from a function generator (Escort EGC 2230) which was connected to the RXHFA input in parallel with a wire antenna. The RXHFA was set to +10 dB gain at 7.050 MHz (daytime) and the frequency of the square wave was 50 Hz.

To do the calibration procedure from blankerinit3.raw, place the file in a directory /demo if you run under Linux or in a directory c:\demo if you run under Windows. Then place a file adfile or adfile.txt in the directory from where you launch Linrad. Use one of these files adfile for Linux or adfile.txt for Windows depending on the OS you are using. You also have to create a subdirectory par under demo. This is where Linrad will place the parameter files for this recording as specified in adfile.

Running for the first time

Press 1 to select blankerinit3.raw as specified in adfile. When running for the first time Linrad will prompt for the receive mode to use. Select any mode, i.e. Normal CW. You must then answer Y to the question Repeat recording endlessly because Linrad will read from the file at 96000 samples per second while you set filter parameters and Linrad will hang in the calibration routine if end of file is reached.

Linrad will remember your choices in the file blnkini_2 under the par subdirectory under demo as specified in adwav. You will then be prompted for the mode parameters unless you already have calibration files for the WSE converters in the directory from where you launch Linrad. (If you have, it might be a good idea to save them and then remove the calibration files (dsp_xxx_corr) from your Linrad directory.

Collect the average pulse shape.

Go to the collect averages screen, look here Calibration for two channels if you are not familiar to the Linrad menu system.

The calibration pulses in blankerinit3.raw are polluted by various 40m signals and that is easily seen on the collect averages screen, figure 1. The strongest signal near the right hand side of the spectrum causes an error in a single FFT bin of nearly 3 dB.



Fig 1. Collecting the average power and phase of 100 Hz calibration pulses polluted with 40 m RF signals.

Set the desired filter response.

When U is pressed on the collect averages screen, Linrad moves to the screen for setting the desired filter response. It looks like figure 2 with the default parameters of Linrad-02.39.



Fig 2. The desired shape screen with interference on the calibration pulses.

The focus on this page is on handling interference with the Refine amplitude and phase correction function, so just press Y on this screen to return to the collect averages screen on which you should then press S followed by X to return to the menu from where you can again go to the calibration menu.

Remove the center discontinuity.

Remove the center notch as the second step of calibrating. The last screen will look like figure 3 if the standard parameters are used. You may look here remove the center notch if you are not familiar to this step.

Press S to save the new calibration function of figure 3.



Fig 3. Here the center discontinuity is removed but discontinuities due to interference are still present.

Refine amplitude and phase correction

The screen of this function looks like figure 4.



Fig 4. The refine screen.

Linrad computes the total energy of the average pulse response and shows a table with average pulse energy outside a range in relation to the total power of the pulse. In this case the total size of the FFT is 2048, but the table shows clearly that there is no reason to use more than 256 points for the pulse. Linrad suggests 128 here, press Y to accept this or N to select another range.

To get an idea what the refine function did to the calibration curves, have a look at figure 5. This is the last screen of the center discontinuity function which is run for a second time. (Answer X this time to not save the result of a second run here.)



Fig 5. After reduction to 128 points, the filter curve is smoother. Compare to figure 3.

Figure 3 does not quite show how bad the original calibration curve is. The number of points is 2048, but the screen is only 900 points wide so some averaging has been done to put the curve on the screen. Figure 5 is not perfect. The interference has caused deviations of up to something like 2% from the correct power. This is an amplitude error of about 0.8 dB. Usually quite harmless, but could be a problem when very long averaging times are used for searching signals far below the noise that change (S+N) with less than 1 dB. The depth of the worst notch in figure 3 is about 3 dB and it is associated with a phase discontinuity that could degrade the blanker performance.

The main purpose of the Refine amplitude and phase function was to reduce the size of the calibration function in old Linrad versions. The FFT size is now limited to the size needed for the time window used to select the pulse which means that the transform size is already reasonably small. This function should only be used in cases where the calibration function contains errors due to interference or poor statistics.

Performance comparisons.

The blankerinit3.raw file can be used to test the performance of the Linrad noise blanker with different calibration functions. Figures 6 and 7 show that there is a noticeable difference, particularly near the ends of the spectrum. Figure 6 uses the noisy calibration function obtained from blankerinit3.raw as described above while figure 7 uses the calibration function from blankerinit2.raw as described here: remove the center notch

At the selected frequency, 7.025 MHz, the difference is about 3 dB as one can see in the S-meter graph.



Fig 6. The smart blanker using the noisy calibration function in 128 points.



Fig 7. The smart blanker using a near perfect calibration function in 2048 points.

The difference between figures 8 and 9 is small, but clearly visible. The dumb blanker removes less than 1% of the data points in both cases but it is more efficient when the pulse shape is closer to ideal as in figure 9.



Fig 8. Both blankers using the noisy calibration function in 128 points.



Fig 9. Both blankes using a near perfect calibration function in 2048 points.