000010 [1]KEYBOARD COMMANDS (available in most modes) A = Numerical display, amplitudes E = Eliminate spur (If AFC present) C = Clear spur elimination tables (If AFC present) G = Save entire screen as .GIF file S = Save input data as file (S again to stop) T = Numerical display, processing delays W = Write output data as .WAV file X = Exit from current mode to menu Z = Restart averaging for S-meter and numerical displays + = Make the wheel mouse step size twice as large. - = Make the wheel mouse step size twice as small. (Note that wheel moves with a pressed wheel acts like '+' or '-') F1 = Help. (Place mouse on a screen object) F2 = Toggle S-meter averaging F3 = Skip D/A output (For fast processing of disk files.) shift F3 = Stop disk file processing (to study morse decode intermediates) F4 = Show RF envelope vs time as an oscilloscope function. ESC = Quit from program [2]Mouse on main spectrum or waterfall graph. Left button selects the frequency for audio output. Right button selects secondary frequency for decode cw to ascii on screen only (not yet implemented) If mouse is on a frequency that is already selected the right button will deselect that frequency. To eliminate a spur, place the mouse cursor on it and press E on the keyboard. To clear spur elimination tables after a frequency change, press C [3]Left button for main spectrum Y-scale expand. [4]Left button for main spectrum Y-scale contract. [5]Left button to move main spectrum upwards. [6]Left button to move main spectrum downwards. [7]Left button to contract main spectrum and waterfall. The lowest frequency is left unchanged if entire spectrum will not fit in window. [8]Left button to expand main spectrum and waterfall. The lowest frequency is left unchanged if entire spectrum will not fit in window. [9]Left button to expand main spectrum and waterfall. The highest frequency is left unchanged if entire spectrum will not fit in window. [10]Left button to contract main spectrum and waterfall. The highest frequency is left unchanged if entire spectrum will not fit in window. [11]Main spectrum and waterfall are calculated as averages of the first fft (fft1). The number of spectra to average over is a multiple of the number in this box. Left button to change. [12]The main spectrum is calculated as an average of the first fft (fft1). The number of spectra to average over is a multiple of the number in this box because averaging is done as averages of averages. Besides affecting cpu load at large numbers of averages this parameter affects the noise blanker because the blanker uses the average according to this number to locate strong signals. The blanker also uses the average of averages to locate strong signals closer to the noise floor. Left button to change. [13]Mouse on high resolution spectrum. Left button to select signal for loudspeaker output. To eliminate a spur, place the mouse cursor on it and press E on the keyboard. [14]Mouse on blanker control bar. This bar shows the noise floor, the noise power level that remains after the blanker has removed pulses. The red vertical line is 20 dB above the quantisation noise of the internal data representation. Make sure your noise floor is not below this level. For setting signal levels, read the Linrad Home Page on the Internet. There are two noise blankers that are intended to be used together. Blue is the colour for the smart blanker while yellow is for the dumb one. When a blanker is enabled, the corresponding control level is indicated with a line on the blanker control bar. In MANUAL mode the control level is fixed and in AUTO mode it is held at some distance above the average noise floor. In both MANUAL and AUTO modes the control level can be changed by grepping the corresponding line with the mouse while the left button is kept pressed. Make sure to not set the control levels too low. The coloured numbers in the upper left and right corners give percentage of points cleared by the blankers. Keep these numbers small, the yellow below 20 and the blue below 4 unless you actually find a better readability of the desired signal with higher numbers. Hint: Use the oscilloscope function to investigate what the blankers do to the noise that causes problems. Also read Linrad Home Page on the Internet. [15]Selective limiter level control for the first fft. Strong signals are prevented from reaching the blankers by the selective limiter. Frequencies that are not routed through the blankers are coloured red in the main spectrum (blue dB scale). Signals that are strong enough to be visible over interference pulses in the main spectrum are found very fast in the first fft. The level is set by clicking the left mouse button in the bar control area. The first fft averaging parameters affect the spectrum as you can see on the screen. Large averaging numbers gives a slower but more accurate determination of strong signals. Look at the spectrum, if parts of the noise floor becomes red one of the limiter controls is too low or some averaging number is too small. [16]Selective limiter level control for the second fft. Strong signals are prevented from reaching the blankers by the selective limiter. Frequencies that are not routed through the blankers are coloured red in the main spectrum (blue dB scale). The second fft is made from the signal after the noise blanker. This spectrum may have much better S/N due to the elimination of interference pulses. It is used to find strong signals that could not be found in the main spectrum because they are below the level of the interference pulses. The second fft averaging parameter "fft2 avgnum" affects the spectrum as you can see on the screen in the high resolution graph (red dB scale). Large averaging numbers gives a slower but more accurate determination of strong signals. Look at the spectrum, if parts of the noise floor becomes red, one of the limiter controls is too low or some averaging number is too small. [17]Button to select mode for the dumb blanker. Toggle between modes by clicking the left mouse button. '-' = OFF 'A' = AUTO 'M' = MANUAL [18]Button to select mode for the smart blanker. Toggle between modes by clicking the left mouse button. '-' = OFF 'A' = AUTO 'M' = MANUAL [19]Button to toggle timf2 oscilloscope. The oscilloscope has a fixed place on the screen. To see it well, move other windows to the right side of the screen. The top track shows power. The next two tracks show I and Q of the "interesting" signal which is obtained from all frequencies that are coloured white in the main spectrum. The two lowest tracks show I and Q of the strong signals, those that are coloured red in the main spectrum. The strong signals are shown at a reduced amplitude, they are affected by an AGC function. The y-scale of the timf2 oscilloscope can not be changed. If you have set the digital signal levels correctly, the scale will fit to the signals you see. Read about "set digital signal levels correctly" on the Linrad Home Page on the Internet. [20]Button to toggle timf2 oscilloscope. The oscilloscope has a fixed place on the screen. To see it well, move other windows to the right side of the screen. The top track shows power summed over both receiver channels. The next four tracks show I and Q of the "interesting" signal which is obtained from all frequencies that are coloured white in the main spectrum. The four lowest tracks show I and Q of the strong signals, those that are coloured red in the main spectrum. The strong signals are shown at a reduced amplitude, they are affected by an AGC function. The y-scale of the timf2 oscilloscope can not be changed. If you have set the digital signal levels correctly, the scale will fit to the signals you see. Read about "set digital signal levels correctly" on the Linrad Home Page on the Internet. [21]Minimum S/N required in the primary AFC search for a new signal. Use the left mouse button to change it. Signals below this threshold do not lock the primary AFC to the frequency. In unlocked mode the primary AFC cursor in the high resolution graph is yellow. In manual mode it will stay on a fixed frequency until the mouse is clicked again to select a signal in the main spectrum (blue dB scale) In automatic mode the AFC will continue to search for a signal above the required S/N value and lock to it as soon as one is found. In the search for a new signal, averages of the main spectra (fft1) are formed for the entire time for which spectra are available. The number of stored spectra is controlled by the parameter "First FFT storage time (s)" in the mode setup. Average spectra are collected under the assumption that the frequency is drifting linearly with time. As a result, many average spectra are searched and the averaging made under the best assumption of frequency drift will give the strongest signal. The maximum frequency drift that will be searched for is set by the parameter "AFC max drift Hz/minute" in the mode setup. The maximum frequency range that the AFC will search over is set by the parameter "AFC lock range Hz". The search range can be reduced by the green bar at the right hand side of this control bar. (For secondary AFC check the "fft3 avgn" parameter box) [22]Minimum S/N required in the search for a new signal. Use the left mouse button to change it. Signals below this threshold do not lock the primary AFC to the frequency. In unlocked mode the primary AFC cursor in the high resolution graph is yellow. In manual mode it will stay on a fixed frequency until the mouse is clicked again to select a signal in the main spectrum or high resolution spectrum (blue or red dB scales) In automatic mode the AFC will continue to search for a signal above the required S/N value and lock to it as soon as one is found. In the search for a new signal, averages of the high resolution spectra (fft2) are formed for the entire time for which spectra are available. The number of stored spectra is controlled by the parameter "Second FFT storage time (s)" in the mode setup. Average spectra are collected under the assumption that the frequency is drifting linearly with time. As a result, many average spectra are searched and the averaging made under the best assumption of frequency drift will give the strongest signal. The maximum frequency drift that will be searched for is set by the parameter "AFC max drift Hz/minute" in the mode setup. The maximum frequency range that the AFC will search over is set by the parameter "AFC lock range Hz". The search range can be reduced by the green bar at the right hand side of this control bar. (For secondary AFC check the "fft3 avgn" parameter box) [23]Primary AFC lock range limitation. Use the left mouse button to change it. Once the primary AFC has been successfully locked to a signal it will try to follow the signal even if it drops below the S/N value required for the initial locking. To prevent drifting away too far when locking to occasional peaks in pure noise in case the desired signal has paused for a while the lock range is limited to half the frequency range selected for the AFC graph. This bar limits the lock range further. (For secondary AFC check the "fft3 avgn" parameter box) [24]Primary AFC search range limitation. Use the left mouse button to change it. The max search range is set by the parameter "AFC lock range Hz" in the mode parameter setup. This bar is used to make the search range smaller in case the desired signal is close to some other (stronger) signal. [25]Click the left mouse button to make AFC graph cover a smaller frequency range and thereby decrease the AFC max lock range [26]Click the left mouse button to make AFC graph cover a larger frequency range and thereby increase the AFC max lock range [27]Click the left button to toggle AFC mode. '-' = AFC disabled 'M' = MANUAL 'A' = AUTO (not yet implemented) The AFC mode affects both the primary and the secondary AFC. (For secondary AFC check the "fft3 avgn" parameter box) [28]Click the left button to toggle timefunction window for spectrum averaging. '-' = No window used. 'W' = Use a sine(t) window. The window makes the primary AFC slower. It is intended to be used together with coherent averaging for extremely weak signals that can not be copied at all with "normal" methods. (For secondary AFC check the "fft3 avgn" parameter box) [29]Number of spectra for the primary AFC to average over while staying locked to a signal. Click in the box with the left mouse button and type in a new value. The averaging is made without any concern of the frequency drift. This number has to be small enough for the peak caused by the signal to be on the same place over the entire period of time for which the averaging takes place. Expand (zoom in) the main spectrum and look at signals that drift with time to learn what the maximum number of spectra to average over is with the mode parameters you have selected. When you know roughly what the maximum value for "fft1 avgn" is for different kinds of signals, then you also know what maximum number to average over in the primary AFC process for similar signals. For extremely weak but very stable signals this number may be large which will allow an accurate frequency locking at the cost of a large time delay. The maximum number possible for this parameter is limited by the "First FFT storage time (s)" parameter in the mode setup. (For secondary AFC check the "fft3 avgn" parameter box) [30]Number of spectra for the primary AFC to average over while staying locked to a signal. Click in the box with the left mouse button and type in a new value. The averaging is made without any concern of the frequency drift. This number has to be small enough for the peak caused by the signal to be on the same place over the entire period of time for which the averaging takes place. Check that the signal is not broadened much in the high resolution graph (red dB scale) when "fft2 avgn" is given the same value. For extremely weak but very stable signals this number may be large which will allow an accurate frequency locking at the cost of a large time delay. The maximum number possible for this parameter is limited by the "First FFT storage time (s)" parameter in the mode setup. (For secondary AFC check the "fft3 avgn" parameter box) [31]Primary AFC time delay (in number of spectra) Click in the box with the left mouse button and type in a new value. The primary AFC process analyzes spectra over a period of time. For weak signals (EME) that only occasionally are visible above the white noise floor, the primary AFC typically will fit the best frequency to 10 seconds of spectrum data. In such a case, best locking is obtained if the interpolated frequency is used, but then the delay caused by the AFC will be 5 seconds. By setting the AFC time delay to a smaller value than the maximum allowed (type in 999 for maximum) it is possible to make the delay smaller. The frequency used in processing will be less accurate, particularly for signals that drift randomly in frequency and are absent completely most of the time. (For secondary AFC check the "fft3 avgn" parameter box) [32]Primary AFC straight line fit (in number of spectra) Click in the box with the left mouse button and type in a new value. The primary AFC associates an average frequency to each spectrum. Each average is based on "Avg" individual spectra which are summed with or without a sine window as set by one of the AFC control boxes. A straight line is fitted to these frequencies and the processing is made under the assumption that the straight line is correctly describing how the frequency changes linearly with time. A longer fit gives a better locking to stable signals and signals that drift linearly with time. (For secondary AFC check the "fft3 avgn" parameter box) [33]Volume control bar. Click the left mouse button in the bar control area. Note that this volume control is different from the volume control on your loudspeaker or volume controls of the soundboard output mixer. Linrad in weak cw mode is designed to run without any AGC. Strong are amplitude limited. Bring this control to maximum to get a saturated signal. Adjust your hardware (loudspeaker and/or soundboard output volumes) for a loud but acceptable maximum sound level. This will be the sound level for saturating signals. When listening to very weak signals, place the volume control bar for the amplitude indicator to the left of it to be placed between 25% and 50% of full scale. The amplitude indicator turns red at the point of saturation and green if the signal is zero (or close to). [34]Mouse on the baseband spectrum (BFO and filter control). These controls are active only within the spectrum display area The upper red vertical line is the BFO frequency in the correct frequency scale. Grab the BFO by placing the mouse curser on it, then press the left mouse button. Then move it while keeping the button pressed. The two other red vertical lines are the BFO placed 10 respectively 100 times closer to the center frequency of the passband. The "zoomed in" BFO controls have to be used when a narrow passband is chosen causing the true frequency of the BFO to come outside the frequency range of the graph. The filter in use is shown in yellow. To change the flat bandwidth, place the mouse in the upper half of the graph anywhere except on the BFO and press the left button. To change the filter steepness, place the mouse in the lower half of the graph anywhere except on the BFO and press the left button. The baseband graph has two indicator numbers at the right side. The upper is the fft3 size in powers of two, the lower is the currently selected baseband processing mode which will change according to the settings in the control boxes. Use this number as your "Default output mode" in the parameter selection to have your favourite processing mode as the default mode. [35]Press the left mouse button to expand the dB scale. [36]Press the left mouse button to contract the dB scale. [37]Press the left mouse button to move the dB scale upwards. [38]Press the left mouse button to move the dB scale downwards. [39]Press the left mouse button to double the third fft size. The current fft3 size is indicated by it's power of two below this box. When the third fft size is doubled, the frequency scale is expanded by a factor of two. The number of data points describing the filter is also doubled which will allow twice as steep filter skirts. The time delay caused by the third fft is also doubled, twice as many data points have to be collected before the transform can be taken. To expand the frequency scale without making the fft size larger, use the buttons in the left side of this window. The spectrum produced by the third fft is used for the secondary AFC in case AFC is enabled. The secondary AFC indicator is the green rectangle in the frequency scale, present only when AFC is enabled. This indicator shows the frequency of the strongest peak found within 50% of the filter bandwidth. The filter is automatically centered on the secondary AFC frequency although the yellow filter curve does not reflect this to save some time for the graphics. (Note that fft3 avgn will affect the secondary AFC) If neither AFC, nor coherent processing is enabled there is no reason to have more than 4 points on the filter. For AFC only, 8 points is enough. When coherent processing is selected 8 to 16 points will be enough for signals strong enough to decode by ear. Coherent averaging is a possibility to decode signals that are some 10dB below "the ear threshold" and the limit for coherent processing is the third fft resolution that should be set to resolve the true bandwidth of the carrier of the CW signal (about 0.2Hz for 144MHz EME). (Coherent averaging is not yet implemented in Linrad) [40]Press the left mouse button to half the third fft size. The current fft3 size is indicated by it's power of two below this box. If you can not reduce fft3 size further, expand the frequency scale with the box in the left upper corner of this window or reduce window size. When the third fft size is reduced, the frequency scale is contracted by a factor of two. The time delay caused by the third fft is also halved. The spectrum produced by the third fft is used for the secondary AFC in case AFC is enabled. The secondary AFC indicator is the green rectangle in the frequency scale, present only when AFC is enabled. This indicator shows the frequency of the strongest peak found within 50% of the filter bandwidth. The filter is automatically centered on the secondary AFC frequency although the yellow filter curve does not reflect this to save some time for the graphics. (Note that fft3 avgn will affect the secondary AFC) If neither AFC, nor coherent processing is enabled there is no reason to have more than 4 points on the filter. For AFC only, 8 points is enough. When coherent processing is selected 8 to 16 points will be enough for signals strong enough to decode by ear. Coherent averaging is a possibility to decode signals that are some 10dB below "the ear threshold" and the limit for coherent processing is the third fft resolution that should be set to resolve the true bandwidth of the carrier of the CW signal (about 0.2Hz for 144MHz EME). (Coherent averaging is not yet implemented in Linrad) [41]Press left mouse button to toggle the baseband oscilloscope on/off. The oscilloscope has a fixed position on screen, move other windows to the right hand side of the screen when using it. The oscilloscope displays the time function (timf3) used to produce the baseband spectrum (fft3). The bandwidth of the signal displayed by the timf3 oscilloscope is set by the parameter "First mixer bandwidth reduction in powers of 2" among the mode setup parameters. This display is intended for use while checking the timf3 noise blanker which is not yet implemented. [42]Press the left mouse button to increase the gain of the baseband oscilloscope. (Only active if the baseband oscilloscope is enabled) [43]Press the left mouse button to decrease the gain of the baseband oscilloscope. (Only active if the baseband oscilloscope is enabled) [44]Press the left mouse button to expand the frequency scale. [45]Press the left mouse button to contract the frequency scale. [46]Mode for audio compression/expansion. "Off" = Normal mode. "Exp" = Amplitude expander on. "Lim" = Simple amplitude clipping. In normal mode the output is identical to the input up to the saturation level. An extremely fast AGC prevents the input to reach above the saturation level which leads to a limitation of the amplitude at the saturation level without the generation of overtones. When the amplitude expander is on, the amplitude variations of the input signal are expanded by an exponential function. The amplitude expander is intended to be used only at bandwidths below 25Hz where the human ear can no longer hear frequency differences causing both signal and noise to be perceived as a tone with constant frequency and varying amplitude. The human hearing is not designed to distinguish well between small amplitude variations of a single tone because of the logarithmic nature of the amplitude response. The exponential expansion compensates for the logarithmic response and makes decoding of extremely weak signals by ear possible for long times without fatigue. The mode setup parameter "Audio expander exponent" can be used to set the curvature of the exponential expansion. The simple amplitude clipping is completely linear up to the saturation level where the signal is limited. Overtones are generated when clipping occurs. The onset of overtone generation can be used to hear small amplitude variations for example when listening to aurora signals. When a signal is much wider than the modulation, the optimum filter should match the signal spectrum. The sound character that comes out does not change between key up and key down since it will be the noise spectrum that fits the filter in both cases. The simple amplitude clipping is an alternative to exponential expansion in such cases. To get the current mode as default, make the mode parameter "Default output mode" equal to the number at the right hand side of this line of boxes. [47]Mode for coherent processing. Use left mouse button to select mode: "Off " = No coherent processing. "Coh1" = "Binaural CW" "Coh2" = Coherent: I and Q to left and right ear respectively. "Coh3" = Full coherent processing with I to both ears. The coherent modes use a second filter that is "Rat " times narrower than the filter shown in yellow on the graph. In binaural CW mode, the narrow filter is typically adapted to the bandwidth of the keying sidebands while the wide filter is 2 to 4 times larger. In coherent modes, the wide filter is adapted to the bandwidth of the keying sidebands while the narrow filter is at least 8 times narrower. (Use button in upper right corner to get more data points on the filter function to allow larger "Rat" values) The Coh2 mode is safe, if the narrow filter fails to follow the carrier due to AFC problems caused by unstable signals or the requirement for small processing delay, no signal at all is lost. If the carrier phase is incorrect, the signal will be present in both ears. This mode is particularly useful for chirping signals. The Coh3 mode is the most sensitive processing mode. It should be combined with AFC with parameters that allow a good locking to the CW carrier. The Coh3 mode gains 3dB S/N by rejecting the noise power that is in the Q channel. There is a further S/N improvement because anything that is in opposite phase to the carrier is also rejected. The Coh3 mode is best used together with "Exp", exponential expansion of the audio amplitude. Particularly when exponential expansion is used it is important to select the wider filter for minimum possible bandwidth. The narrower filter which is controlled by the Rat parameter must be wide enough to allow the full bandwidth of the carrier to pass. To get the current mode as default, make the mode parameter "Default output mode" equal to the number at the right hand side of this line of boxes. [48]Press the left mouse button to toggle delay between ears. This mode can be enabled only when "Coh" or "X+Y" is not selected. This mode will delay the signal to one ear, thus creating a frequency depending phase shift between the two ears. The effect of the phase shift is an artificial stereo effect. Different frequencies seem to arrive from different spatial directions. The "Del" mode may be useful when the selected bandwidth is larger than the bandwidth of the desired signal which can be useful for unstable signals or when a very small processing delay not allowing AFC is required. To get the current mode as default, make the mode parameter "Default output mode" equal to the number at the right hand side of this line of boxes. [49]Press the left mouse key to toggle "X+Y" mode. This mode can be enabled only when "Coh" is not selected. It will deselect "Del" automatically. The "X+Y" mode routes the two receive channels to the two output channels for stereo reception. This mode is useful when polarisation changes too quickly to allow the automatic polarisation adaptation to follow. The signals for the output is two orthogonal combinations of the two input signals. Set the polarisation to "Fixed" and horizontal to receive horizontal in one ear and vertical in the other. To get the current mode as default, make the mode parameter "Default output mode" equal to the number at the right hand side of this line of boxes. [50]Click the left button to force two output channels. Can be used only when the mode does not require two output channels. Modes that require only a single output channel puts the same signal to both ears. If two output channels are forced, the signal is twisted by 180 degrees between the ears. Twisting the phase may relieve fatigue after long periods of listening to weak signals. To get the current mode as default, make the mode parameter "Default output mode" equal to the number at the right hand side of this line of boxes. [51]Press left button to toggle output between 8 and 16 bits. Not really useful - if your hardware allows 16 bit, use them all. Could perhaps reduce processor load on very slow computers(?) To get the current mode as default, make the mode parameter "Default output mode" equal to the number at the right hand side of this line of boxes. [52]Press left button to change "Rat". Enter new value from keyboard. Rat is the ratio of the bandwidths for the two filters used in coherent modes. To get the current mode as default, make the mode parameter "Default output mode" equal to the number at the right hand side of this line of boxes. [53]Press left button to change the delay between ears in "Del" mode. Enter new value from keyboard. To get the current mode as default, make the mode parameter "Default output mode" equal to the number at the right hand side of this line of boxes. [54]Press left button to toggle between "Fixed" and "Adapt". In "Fixed" mode the two incoming signals are combined to produce one output signal that correspond to the polarisation shown above which is further processed and shown in green in the high resolution graph and in the baseband graph. The red spectra in the graphs is the orthogonal polarisation. In "Adapt" mode Linrad uses available information to adjust the polarisation to fit the polarisation of the incoming wave. Ideally all signal should be present in the green spectra while no signal (20dB less) should be visible in the red spectra. [55]Press left button to change polarisation angle. This function is intended for use in "Fixed" mode. [56]Press left button to change ellipticity of polarisation. When the control line is at the extreme left or the extreme right within the box, the phase shift between the two incoming signals is 90 degrees. Then the resulting polarisation can be set to circular, left or right hand rotation by changing the angle in the blue field above. When this control is at it's center position the polarisation is linear. Together angle and ellipticity can be set to match any polarisation, linear, circular or elliptic. [57]Press left button and move vertical line to change time constant of polarisation evaluation. Only available in "Adapt" mode. For extremely weak signals with stable polarisation such as 144MHz EME, use a long time constant. Signals that rapidly change polarisation like 7MHz at night need use a short time constant. Success of adaptive polarisation is indicated by how well the signal is suppressed in the red spectra in the high resolution graph and in the baseband graph. [58]Parameter "fft1 avgnum". Press left mouse button and type in new value. This parameter controls the averaging of the main spectrum, fft1, which has blue lines for the dB scale. Larger values give higher sensitivity but slower response. The parameter "fft1 avgnum" is always a multiple of the primary average number which can be changed by the box in the lower right corner of the main spectrum. When the first fft bandwidth is high, the averaging process becomes a substantial part of the total computing and averaging in groups of up to 5 spectra saves time. [59]Parameter "fft1 avgnum". Press left mouse button and type in new value. This parameter controls the averaging of the main spectrum, fft1, which has blue lines for the dB scale. Larger values give higher sensitivity but slower response. The parameter "fft1 avgnum" is always a multiple of the primary average number which can be changed by the box in the lower right corner of the main spectrum. When the first fft bandwidth is high, the averaging process becomes a substantial part of the total computing and averaging in groups of up to 5 spectra saves time. The averaged fft1 spectrum is used for the selective limiter. The sensitivity/speed compromise affects the noise blanker operation and the blue control bar at the left side of the high resolution graph must be set in accordance with the fft1 avgnum value selected. [60]Parameter "waterfall avgnum". Press left mouse button and type in new value. This parameter controls the averaging for each line of the waterfall graph. Larger values give higher sensitivity but slower response. The parameter "waterfall avgnum" is always a multiple of the primary average number which can be changed by the box in the lower right corner of the main spectrum. When the first fft bandwidth is high, the averaging process becomes a substantial part of the total computing and averaging in groups of up to 5 spectra saves time. [61]Parameter "waterfall avgnum". Press left mouse button and type in new value. This parameter controls the averaging for each line of the waterfall graph. Larger values give higher sensitivity but slower response. [62]Parameter "fft2 avgnum". Press left mouse button and type in new value. This parameter controls the averaging of the high resolution spectrum, fft2, which has red lines for the dB scale. Larger values give higher sensitivity but slower response. [63]Parameter "fft3 avgnum". Press left mouse button and type in new value. This parameter controls the averaging of the baseband spectrum, fft3, which has green lines for the dB scale. Larger values give higher sensitivity but slower response. The averaged fft3 spectrum is used for the secondary AFC if AFC is enabled. The secondary AFC locates the maximum within 50% of the selected bandwidth and places the secondary AFC cursor at the peak position. The secondary AFC indicator is the green rectangle in the frequency scale of the baseband graph, present only when AFC is enabled. The filter is automatically centered on the secondary AFC frequency although the yellow filter curve does not reflect this to save some time for the graphics. Note that the value selected for "fft3 avgn" directly affects the secondary AFC. [64]Parameter "Waterfall zero". Press left mouse button and type in new value. This parameter selects the zero point (in dB) for the waterfall graph. [65]Parameter "Waterfall zero". Press left mouse button and type in new value. This parameter selects the zero point (in dB) for the waterfall graph. It will also affect the vertical position of the high resolution graph (red dB scale). The colour scale currently in use for the waterfall is displayed at the right hand side of the high resolution graph. [66]Parameter "Waterfall gain". Press left mouse button and type in new value. This parameter selects the gain for the waterfall graph. [67]Parameter "Waterfall gain". Press left mouse button and type in new value. This parameter selects the gain for the waterfall graph. It will also affect the dB scale of the high resolution graph (red dB scale). The colour scale currently in use for the waterfall is displayed at the right hand side of the high resolution graph graph. [68]The three S-meters in the coherent graph show the strength of the signal that has passed the selected baseband filter. Set the filter to rectangular with 1kHz bandwidth to get the noise floor in 1000Hz. Then add 30 dB to get dB/Hz. The three meters show: Top = Peak S-meter. Middle = Current S-meter value. Bottom = True RMS for everything that passes the filter. The S-meter shows the level of a CW signal that has a speed that fits the selected passband. It is obtained from a bi-directional fast attack, slow release amplitude follower. The S-meter gives a level for CW signals that is independent of the mark to space ratio of the received signal. By pressing F2 you can get averaged true RMS readings above the normal three S-meter readings. Clear and restart with F2. The top line is the number of averaged values. [69]Click this box to toggle the coherent oscilloscope on/off. The coherent graph oscilloscope shows various signals in the baseband. These oscilloscope images are intended to assist in the development of automatic detect algorithms for CW and various digital modes. For details, check the code in coherent.c [70]Click in this box to open or close the eme window. [71]Use this box to enter the ww locator of another station to get the optimum tx polarisation and some more info about the dx location. [72]Use this box to enter a call sign or fragments thereof. Linrad will search it's database and give the optimum tx polarisation and some more info about the dx location in case the search gives a single station and the database contains the geographical location. [73]Timf2 oscilloscope: Icrease gain for strong signals [74]Timf2 oscilloscope: Decrease gain for strong signals [75]Timf2 oscilloscope: Icrease gain for weak signals [76]Timf2 oscilloscope: Decrease gain for weak signals [77]Timf2 oscilloscope: Toggle graphical modes P=Change to set pixels L=Change to draw lines [78]Timf2 oscilloscope: Toggle graphical modes H=Change to hold C=Clear and change to normal mode [79]Use this box to toggle AGC on/off. [80]Use this box to enter time constant for AGC attack. A long time constant will prevent the AGC from reducing the gain due to short bursts of interference. This parameter will affect the release time which can never be smaller than the attack time. [81]Use this box to enter time constant for AGC release. [82]Click this box to decrease the center frequency for your receiver hardware. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [83]Click this box to increase the center frequency for your receiver hardware. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [84]Click this box, then enter a new center frequency for your receiver hardware. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [85]Click this box to decrease the gain of your hardware. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [86]Click this box to increase the gain of your hardware. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [87]Click this box, then enter a new gain value for your hardware. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [88]Parameter bg.afc2_range The second AFC is enabled if the first AFC is enabled and/or if coherent processing is enabled. The range parameter limits the range covered by the second AFC. Set this parameter to zero to disable the second AFC unconditionally. The second AFC uses the spectrum shown in the baseband window, (this window) and the amount of averaging affects the second AFC. This parameter bg.afc2_range can be set from 0 to 99. The maximum range corresponds to 50% of the baseband bandwidth. [89]Parameter bg.afc2_delay The second AFC is enabled if the first AFC is enabled and/or if coherent processing is enabled. It can be disabled by setting the range to zero. The second AFC uses the spectrum shown in the baseband window, (this window) and the amount of averaging affects the second AFC. This parameter bg.afc2_delay determines the delay associated with the second AFC [90]Press left button. Then type a new calibration constant in dB. The dBm scale and the S-Units scale have separate calibration constants. The dB scale is the same as for the power meters in the coherent graph. [91]Press left button to toggle scale between dB, dBm and S-Units. [92]Press left button to increase the averaging in the meter graph. The averaging affects both the curves and the bar graph. [93]Press left button to decrease the averaging in the meter graph. The averaging affects both the curves and the bar graph. [94]Press left button to expand the vertical scale. [95]Press left button to contract the vertical scale. [96]Press left button to move graph downwards. [97]Press left button to move graph upwards. [98]Press left button to toggle curves between: P = Peak power. M = RMS power. 2 = Both RMS and peak power. Note that both the selected polarisation and the orthogonal polarisation is shown (in different colours) when two RF signals are fed into Linrad. The S-Meter bar shows only the selected polarisation just like the power meter in the coherent graph. [99]Click this box to show/hide the S-meter graph. [100]Mouse on a border line. Press left button to resize window (and/or move window) [101]Mouse on a border line. Press left button to move window. [102]Mouse on the S-Meter border line. Press left button to resize window (and/or move window) The S-Meter is a bar graph when the window width is small. Set the window more than 13 characters wide to get curves showing signal level vs time. When the window is made wide, control buttons are present. Use these buttons to set the graphs to show RMS power, peak power or both and to select scale in dB, dBm or S-Units. The bar graph is always peak power in S-Units to conform to conventional S-Meters. [103]Click this box to decrease the transmit frequency. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [104]Click this box to increase the transmit frequency. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [105]Click this box, then enter a new transmit frequency. The default routine will fit the WSE converters. Read the file z_USERS_HWARE to find out how this control can be configured to fit your own hardware in case you use something else. [106]Click this box, then enter a number to select the speech processing parameters to use for transmitting. Only numbers actually set up and saved are available. Use 'V' in the main menu, then 'C' to get to the speech processor set-up routine or make files par_ssbprocXX manually by making changes to existing files that you renumber. XX is a number 0 to MAX_SSBPROC_FILES (defined in tx.c) [200]For information about mode parameters, place curser on the corresponding line and press F1. The mode parameters are intended to be set up once and for all in accordance with the abilities of your hardware. One set of mode parameters can be selected for each processing mode. [201]The first fft bandwidth is used to set the size of the first fft to a bandwidth within a factor of two of the desired value. If the second fft is enabled, there is no reason to select a very narrow bandwidth since the first fft then is used only for the selective limiter that prevents strong signals from reaching the noise blanker. When second fft is disabled the first fft bandwidth will directly affect the sensitivity of the waterfall graph and it will also affect AFC performance. Narrow bandwidths cause appreciable processing delay, this is unavoidable since data has to be collected over a time period that is about twice the time given by 1/bw, (depends on what window is selected) [202]The window function used for the first fft is a power of sin(t). Low powers give faster processing but the filter function associated with each frequency bin gets less steep skirts. When a large bandwidth is selected for fft1, higher window powers are required to produce enough attenuation a few hundred Hz away from very strong signals. Inject a very strong carrier into the antenna input and look at the main spectrum. The width of the peak you see gives an indication of the frequency band over which a strong signal will cause spurs when listening to a weak signals. For a fft1 bandwidth of 200Hz the window power should be 3 or more. At bandwidths below 10Hz a window power of 1 is sufficient. As a measurement tool for close in sideband noise window powers up to 9 may be useful. [203]Version number for fft1 routine. Depending on your hardware there are several different fft implementations available. Check the timing of each one of them and select the fastest if there is a significant difference. Note that the mode 2 is better to use in case you will not calibrate your system with a pulse generator since all the other modes use a transformation from real to complex format that introduces an increased sensitivity at lower frequencies which needs the calibration routine to get eliminated. [204]First fft storage time. This parameter affects memory allocation for old fft1 transforms. It affects the maximum number of averages you may use for the main spectrum. In case second fft is deselected this parameter also affects the memory allocation for AFC and spur rejection in case these functions are enabled. [205]First fft gain. If you have disabled the second fft, this parameter will just shift the dB scales just like a volume control. In case you use the second fft you should use this parameter to set the noise floor at the input to the first backwards fft. Press A on the keyboard while your system is running to get amplitude information in the lower left corner. "Floor" is the number of bits RMS for the noise floor of the signal entering the first backwards fft. A larger value for gain here will increase "Floor". Your loss of system noise figure because of quantization noise will be: Floor NF loss (RMS voltage) (dB) 1 0.4 2 0.2 4 0.1 10 0.04 Placing the noise floor too high may lead to saturation in later processing steps. Use "First FFT gain N" to set the noise floor as high as possible. Then use this parameter to place the floor somewhere between 3 and 30 according to your preferences. [206]The second fft is the high resolution fft that is intended for use at large bandwidths (20kHz analog bandwidth and more). If your analog hardware is a conventional radio with a few kHz bandwidth there is usually no reason to enable the second fft. Only if your radio has very low distortion for signals within the passband and you are troubled by impulse noise that can not be removed by the noise blanker of your radio because of the presence of strong signals at nearby frequencies the second fft will be useful since it allows noise blanking in the presence of strong signals. [207]Version number for back transformation from fft1 to timf2. Depending on your hardware there may be several different fft implementations available. Check the timing of each one of them and select the fastest. If your computer supports MMX instructions the differences may be substantial. [208]Sellim maxlevel This parameter sets the maximum level for strong signals. A lower value here will allow more gain in the second fft without risk of overflows. Press A on the keyboard while your system is running to get amplitude information in the lower left corner. If "fft1 St" , "timf2 St" or "fft2" overflows (margin becomes zero) you may try to decrease this parameter. A small value here may make desired signals distorted because they are classified as "strong" and become compressed in a way that degrades S/N. [209]Number of zero gain butterfly loops for the first backwards fft from fft1 to timf2. What value to choose depends on your hardware and the fft1 parameters you have selected. Press A on the keyboard while your system is running to get amplitude information in the lower left corner. Make the "First backward FFT att. N" as small as possible, but make sure the amplitude margin of timf2 Wk and Timf2 St does not become zero. Occasional saturation of timf2 St is ok but timf2 Wk should never saturate. Read about "set digital signal levels correctly" on the Linrad Home Page on the Internet. [210]Set the bandwidth of the second fft. The second fft size can be set equal to or larger than the first fft. Very narrow bandwidths cause some processing delay but give enhanced sensitivity for very weak CW signals. Narrow bandwidths are required for the AFC to lock properly to extremely weak signals. [211]The window function used for the second fft is a power of sin(t). Low powers give faster processing but the filter function associated with each frequency bin gets less steep skirts. When a large bandwidth is selected for fft2, higher window powers are required to produce enough attenuation a few hundred Hz away from very strong signals. Since very strong signals are already removed by the selective limiter associated with the first fft there is no reason to use powers above 2. At narrow bandwidths, a few Hz and below it is perfectly ok to skip the window completely and use 0 for this parameter to save processing time. [212]Version number for the second fft. Depending on your hardware there may be several different fft implementations available. Check the timing of each one of them and select the fastest. If your computer supports MMX instructions the differences may be substantial. [213]Number of zero gain butterfly loops for the second fft. What value to choose depends on your hardware, the fft1 and the fft2 parameters you have selected. Press T on the keyboard while your system is running to get amplitude information in the lower left corner. Make the "Second forward FFT att. N" as small as possible, but make sure the amplitude margin of fft2 does never become zero. Read about "set digital signal levels correctly" on the Linrad Home Page on the Internet. [214]This parameter reserves memory for storing old fft2 transforms. Old transforms are used by the AFC. For tracking really weak signals you may need 30 seconds here. Old transforms are also used to recalculate average fft2 spectra when the frequency is changed. If you can not set large enough numbers for fft2 avgnum, make this parameter larger. Without AFC a typical fft2 storage time is 4 seconds. [215]If you enable the AFC, make sure you have allowed enough storage for old transforms, "First FFT storage time" or "Second FFT storage time". [216]Max AFC search range. This parameter and the AFC max drift parameter determine how much memory is allocated for the AFC routine. In case you select a very narrow bandwidth for the fft's used for the AFC in combination with a very large search range the time to lock to a signal may become excessive. Do not make the search range much larger than required to accommodate for the uncertainty in placing the mouse on a signal in the graphs. If very high resolution is selected for the second fft, use the high resolution graph to lock to the desired signal. [217]Max frequency drift for signal search. The AFC searches for signals that drift linearly with time. Averages of all the old transforms are made under assumption of all possible frequency drifts within the limits given by this parameter. This parameter in combination with the search range parameter determines memory allocation and time consumption of the search process. [218]Enable automatic morse code detection. The morse code transmission from the selected signal is translated into ascii and presented on the screen. (This function is not in place yet, some code is present in coherent.c but it is far from complete in linrad00-44 / July 2002) [219]Spur removal is a PLL loop that can lock on very stable signals and create a carrier in the opposite phase with the same amplitude. That carrier is then subtracted to remove the spur. The effect is a very deep and very narrow notch filter. The process is efficient in terms of cpu usage so many spurs can be treated simultaneously. Spur removal operates in the frequency domain using the second fft if available, otherwise it uses the first fft. The notch will always be narrower than the fft resolution. If your spurs are somewhat unstable, increasing the fft bandwidth may help. [220]The spur time constant is used to determine from how many transforms the amplitude and phase of the spur should be calculated. The fft2 storage time (or fft1 if fft2 is not present) has to be long enough to keep the transforms in memory. [221]First mixer bandwidth reduction. The first fft and the second fft (if enabled) produce data at the data rate of the A/D board. When a frequency is selected the signal is mixed with an internal digital oscillator of the same frequency to shift the desired signal to zero frequency (baseband I and Q). The signal is then low pass filtered to remove everything above some threshold frequency. When no signal is present above the threshold frequency it is possible to reduce the data rate (sampling speed) without introduction of alias spurs. This parameter determines the maximum bandwidth available for the baseband processing. It also sets the sampling speed for the corresponding time function timf3. The baseband noise limiter (not yet implemented) operates on the timf3 without any selective limiter so it will work only if the baseband bandwidth is set small enough to exclude strong undesired signals. [222]First mixer no of channels. It is possible to select secondary channels by use of the right mouse button. These channels are intended to be processed in the same way as the main channel but they will not be routed to the loudspeaker. The secondary channels are intended to be useful when the automatic implementation of morse code to text is in place. The idea is to allow text information on screen to keep track of what other stations work. The automatic cw to text translation is not expected to be as good as a trained human operator but it will surely be good enough to allow a good overview of what other stations work. [223]Baseband storage time. Time span over which the filtered baseband signal is saved. This is the time over which coherent averaging will search for matching keying sequencies. (Some day hopefully....) [224]Output delay margin Set this parameter to a small value to get a small processing delay Press T to get timing information while receiving. If the D/A MIN value becomes zero the selected value is too small. [225]Output sampling speed for the D/A. Some audioboards allow different sampling speed for A/D and D/A. In case very high A/D bandwidths are used linrad may need two audioboards to allow a reasonable sampling rate for the output. For morse code there is never any reason to set the output speed above 5kHz. For voice modes it may be useful to set sampling speeds up to 10kHz. The output routines are not optimised for high speeds, they use trigonometric functions and other slow processes that are perfectly ok at 5kHz even on slow computers but that cause a severe load even on a PentiumIII at 44.1kHz. The processor load increases rapidly with the output sampling speed. [226]The default output mode sets the combination of selections in the line of boxes in the baseband graph to their an value. Set this parameter to the value at the right hand side of the line of boxes that you find with your own favourite settings. [227]The amplitude expander exponent is used to determine the curvature of the expander when the "Exp" option is selected in the baseband graph. [228]Messages 200 to 299 reserved for parameter menu. [300]REMOVE CENTER DISCONTINUITY When I and Q are sampled by means of a soundcard the very low frequencies are missing because soundcards are AC coupled. For this reason there is an infinitely deep notch in the spectrum at the frequency corresponding to a DC voltage in I and Q. This means that the spectrum collected as the average frequency response is actually two spectra separated by the notch. The amplitude is zero and the phase may shift by an arbitrary angle across the notch (the center discontinuity) in the measured data. The notch may be much wider than one would expect from the hardware characteristics. The spectrum is measured by repetitive pulses but each pulse is processed separately. This means that a pulse is processed with only as many data points surrounding it as permitted by the pulse repetition frequency of the calibration pulse generator. If the PRF is 100 Hz, each pulse is processed with data points covering a time span of about 50 milliseconds. The lowest frequency the pulse may contain is therefore 200 Hz. Cutting the time function like this produces a high pass filter with an oscillatory behaviour in the frequency domain and this high pass filter adds to the notch produced by the hardware to produce a wider notch with oscillatory behaviour. The purpose of this routine is to find the correct phase and amplitude across the notch. This is done by splitting phase and amplitude into their symmetric and the asymmetric parts. Then polynomials are fitted to the measured data and the polynomials are used to establish phase and amplitude across the notch. The green vertical line marks the range up to where the polynomial is fitted. (press B to get to the screen where it can be changed) Place the green line to the left of the structures attributable to the hardware and the PRF of the calibration pulses. The limit frequency is typically ten times the PRF or five times the 3dB point of the hardware HP filter. The yellow curve is the measured data. Use + and - to see how it behaves. The red curve is the difference between the measured data and the polynomial fitted with the parameters given for ch0 (and ch1 in a two channel system). Select the smallest number of terms that places the red line close to the x-axis in the region to the left of the green line. Move the blue line to a suitable position where the red line is on the x-axis and parallel to it. When you press A, the calibration curve will be constructed from the yellow curve at the left side of the blue line and from the polynomial at the right side of the blue line. If the blue curve were placed at a point where the red line is significantly displaced from the x-axis the new calibration curve would have a step corresponding to the displacement at the point of the blue line. Try it and repeat the center discontinuity removal routine to get an understanding for how it works. Avoid it when finally calibrating your receiver. [301]INITIALIZE THE EME DATABASE If you did it before and do not want to change anything, press enter or any other key except M to use your old file: /home/emedir/linrad_dxdata (Linux) C:\emedir\linrad_dxdata (Windows) ************ In order to decide what polarisation to use for transmit Linrad has to know the geographical location of the other station. There are files on the Internet containing this information. Make a directory /home/emedir for Linux or C:\emedir for Windows and copy the data files there so Linrad can find at least one of these files: /home/emedir/allcalls.dta or C:\emedir\allcalls.dta /home/emedir/eme.dta or C:\emedir\eme.dta /home/emedir/dir.skd or C:\emedir\dir.skd /home/emedir/CALL3.TXT or C:\emedir\CALL3.TXT (Search for emedir.exe allcalls.dta, CALL3.TXT and vhfsched.skd on the Internet) Linrad will create the file own_info, a file containing your own location, UTC offset and whether the EME database should always be installed so you will not have to press M on the main menu. The database will occupy some memory and you may want to use memory for other purposes now and then on a small computer, that is why you have an option here. Linrad will also create linrad_dxdata which contains the data from all the files in Linrad's own format. If the files contain incompatible information, the file location_errors.txt will report the errors. All files are created in /home/emedir(Linux) or C:\emedir(Windows) Once the EME database is initialized, the EME window can be used for finding call signs by a search for fragments using ? and * as "wildcards" [302]ROUTINES TO CALIBRATE LINRAD TO FIT YOUR HARDWARE IN I/Q MODE. In I/Q mode you have two mixers that operate in quadrature to produce two audio signals for your computer (for each RF channel). Ideally the amplitude of I and Q should be equal and the phase shift should be 90 degrees exactly between the signals. In real life component tolerances lead to errors which to some extent can be minimized by adjusting the hardware. In case the hardware contains filters it is very difficult to make them equal in two channels and if the filters are steep as in the WSE RX2500 it is practically impossible. Linrad compensates the errors in the balance between I and Q in the frequency domain. This way the balancing parameters are functions of the frequency and allow a good image rejection over the entire bandwidth. Calibrating I/Q phase and amplitude should be the first step of your calibration. In I/Q mode the amplitude and phase characteristics are usually quite good but Linrad allows you to make it perfect. The calibration steps B, C, and D have to be performed in this order. C uses the output from B and D uses the output from C. [303]ROUTINES TO CALIBRATE LINRAD TO FIT YOUR HARDWARE. You are running linrad with a single audio channel for each radio channel. This means that an IF filter is used to remove the mirror image from the last frequency conversion in your hardware. Typically steep filters are used in this mode and steep filters always have an oscillatory impulse response and also often an amplitude vs frequency curve with some ripple. The calibration procedure allows you to correct the frequency and phase responses of your entire system. Linrad will find the optimum characteristics for a digital filter that will be added in your signal path and that will give you a perfectly flat frequency response and an optimum impulse response. The first step A has to be performed first. B uses the output of A. [304]CHECK YOUR PULSE GENERATOR AND SET AMPLITUDE ADN PRF CORRECTLY. For the calibration to become successful your pulse generator should give strong pulses while not sending noise into your hardware. On this screen you can see the S/N of your pulses as the peak power of a typical pulse in relation to the RMS noise floor. A good wideband system can give S/N above 70 dB and allows the usage of very low pulse repetition frequencies, down to 10 Hz. The S/N in the next step when the pulses are actually used is determined by the total energy of a single pulse in relation to the total energy collected in the time interval between two pulses. PRF S/N Level S/N second step Quality 10Hz 75dB 30% 37dB Perfect 10Hz 60dB 3% 25dB OK 10Hz 50dB 1% 16dB Useless 75Hz 60dB 3% 34dB Perfect 75Hz 50dB 1% 25dB OK 75Hz 40dB 0.3% 15dB OK (Ignore) 200Hz 40dB 0.3% 20dB OK 200Hz 30dB 0.1% 12dB Questionable (Ignore) In this table (Ignore) means you have to press 'I' to ignore the error "low S/N". The disadvantage with using a high PRF is that the center discontinuity becomes wider because the lowest frequency seen by the calibration routine must have a full period in the time interval between two pulses. A wide discontinuity at the center could make "Remove center discontinuity" more difficult. One would have to interpolate over a wider frequency range. Make sure to use a PRF low enough to make the noise floor flat between each pulse. Note that the calibration uses averaging and that it is affected by pulse oscillations well below what you can see on this screen. [305]COLLECT THE AVERAGE PULSE RESPONSE OF YOUR SYSTEM The magenta trace is the amplitude and the green trace is the phase. If you can see narrowband signals in the amplitude curve there is a risk that you can get discontinuities in the phase function. Always look for a while and make sure it does not happen. The cure to such problems is better screening, a different frequency of operation or stronger calibration pulses. The pulse response you see on screen is your hardware plus whatever calibration function you may have in RAM. Try to clear RAM in case you see odd results. Whatever you do will be harmless as long as you do not save on the hard disk with the 'S' command. [306]MANIPULATE THE AVERAGE FREQUENCY RESPONSE On this screen you are given the opportunity to remove non-valid data from the average frequency/phase vs frequency curves that Linrad has collected. Normally there is no reason to use these functions, the data collected can usually be used as they are with no manual adjustment. Just press 'A' to use data as they are. [307]SELECT THE DESIRED FREQUENCY RESPONSE The white curve is the frequency response that you will force the entire system to have. The yellow curve is the amplitude vs frequency function that Linrad will have to use to achieve this goal. Note that it is impossible to ask for a flat frequency response that extends to frequencies where Linrad has no information. You can not have a flat response at frequencies where your hardware does not allow any signal to pass. Be careful to make the yellow curve reasonably flat. Use the keys A to F to make the white curve compatible to your hardware. The goal is to make a white curve that "looks nice" but with an associated yellow curve that does not have large peaks. At frequencies where your hardware provides high attenuation Linrad will have to apply high gain to make the total response flat. The maximum gain needed (relative to the mid-band gain) is given on screen. Do not allow this number to go above 10 dB. Look for "Max correction" do not allow it to be above 10 dB. Very steep skirts can be set, but the impulse response will become very long if you select extreme shape factors. If you have no QRN problems it might be nice to have an extremely rectangular filter response, but the noise blanker could perhaps be affected in a non-desirable way. When this text was written (Sept 2006) it was not known what kind of total filter response would be optimum under different circumstances. Maybe it does not matter??? [308]REMOVE CENTER DISCONTINUITY Linrad has no information about phase and frequency response for audio frequencies close to zero. Soundcards are typically not DC-coupled... Other hardware such as the SDR-14 may be DC-coupled, but Linrad does not have any information about the DC levels or frequencies below the repetition frequency of your pulse generator. Due to missing information the phase response has a discontinuity at the center frequency. The amplitude may also be low and this could be because of a high pass filter in a soundcard and/or due to the high pass filter that Linrad uses to pick single pulses. Due to peculiarities in the digital filters inside soundcards that remove the DC component, it is necessary to apply a window function in the time domain around each pulse because the DC level is artificially shifted by rounding errors so it is different on both sides of a pulse. This is the origin of the high pass filter inside the Linrad calibration routine. The purpose of this function is to interpolate across the center discontinuity. This is done by a polynomial fitting across the discontinuity while excluding the points close to the discontinuity. The fitted curve is then used to replace the collected data near the discontinuity. On this screen you have to set a suitable width of the region where fitted data is to be used. Subsequent screens will show if you set the limits too wide or too narrow. [309]APPLY THE FITTED FUNCTIONS ACROSS THE DISCONTINUITY OR LEAVE ORIGINAL DATA UNCHANGED If you press S here, the interpolated data will replace your measured data in the dsp_xxx_corr files. You may try to exclude a wider frequency range in a subsequent attempt, but since the original data is lost it will be impossible to try to fit with a narrower frequency range excluded. [310] REFINE CALIBRATION FUNCTION The calibration function is amplitude and phase in the frequency domain - or it is the averaged pulse response in the time domain. During calibration the size of the FFTs used are typically far too large which is the same as to say that the pulse response contains a too long range of noise only, surrounding the averaged pulse. At this point Linrad has decided how many points the pulse actually needs. If you allow save in a smaller size than the original, Linrad will smooth the noise floor surrounding the averaged pulse. In case your calibration function contains some spurs (peaks in the frequency domain) Linrad will try to remove those peaks. It may be a good idea to save the calibration function before saving a smaller calibration function in case you spent a lot of time collecting it. Reducing the size of the calibration function will usually provide a flatter noise floor - but this process can go wrong and lead to a less flat noise floor. ----- Warning!! ----- Reduce the number of points in a single run of this function. If a calibration function that is not in its original size is run through this procedure the result will not be quite correct. [311]BALANCE AMPLITUDE AND PHASE BETWEEN I AND Q In direct conversion mode two audio channels from each RF channel are used. The amplitude and phase of these channels has to be very carefully balanced to avoid false signals at the mirror image frequency. If you set the center frequency to 7.050 Mhz and click on 7.040 MHz to receive a signal there, a direct conversion receiver will also receive signals at 7.060 MHz - the mirror image frequency. Rather than requiring the user to tweak his hardware very carefully, Linrad measures the level of the mirror image at many frequencies and sets up a function that will be used to correct for hardware errors. Making the correction in the frequency domain is advantageous because the two audio channels I and Q may then have different frequency responses and time delays. It can be VERY difficult to obtain a reasonable mirror rejection over the entire passband when tweaking the hardware for I and Q to have identical frequency responses. If you have two RX channels you must calibrate both of them simultaneously. [312]The Linrad S-meter can show Peak power or RMS power in S-units, dB or dBm. Make the window small in the X-direction to see signal strength as a bar graph or make the window wide to see level vs time curves. Use the F1 help on the different boxes that appear when the graph is made wide to find out how to set the window to show what you want to see. With two RF channels, the S-meter shows the signal level from the same two orthogonal combinations as shown in the baseband graph. Peak power is GREEN and MAGENTA corresponding to the two curves in the baseband graph just as WHITE is baseband spectrum and S-meter peak power for a single channel. The RMS powers corresponding to each peak power are BLUE/GREEN, RED,MAGENTA and YELLOW/WHITE. [313] adwav missing. To use .wav files as the input to Linrad you must have a text file named adwav in the logged directory from where Linrad was started. For each .wav file you should make a single line in adwav starting with the full name of the .wav file. Optionally you may add a second full filename to specify a group of parameter files for this particular .wav file so each file can have its own parameters. Example adwav (linux): /wavlib/unkn422.wav /wavlib/parfiles/par_unkn422 Example adwav (Windows): C:\wavlib\unkn422.wav C:\wavlib\parfiles\par_unkn422 Note that all directories must exist. E.g. in this example the directory "parfiles" must exist before you start Linrad. Linrad will never create directories. [314] adfile missing. To use raw data files as the input to Linrad you must have a text file named adfile in the logged directory from where Linrad was started. The raw data files have to be saved by Linrad. They will contain the calibration information from the system on which they were recorded as well as time and frequency. (Create raw data files by pressing 'S' while Linrad is running) For each raw data file you should make a single line in adfile starting with the full name of the raw data file. Optionally you may add a second full filename to specify a group of parameter files for this particular raw data file so each file can have its own parameters. Example adfile (linux): /rawlib/myfile /rawlib/parfiles/par_myfile Example adfile (Windows): C:\rawlib\myfile C:\rawlib\parfiles\par_myfile Note that all directories must exist. E.g. in this example the directory "parfiles" must exist before you start Linrad. Linrad will never create directories. [315] SETUP for the Linrad network Receiver input can be the hardware connected to this computer or a signal received from the network. Raw data input from the network can not be combined with raw data output to the network. The fft1, timf2 and fft2 outputs can be enabled on a computer that uses the network for input while the fft1 input can be combined only with the timf2 and fft2 outputs. Pick a base port and a send address for the master computer, the one that is connected to the radio hardware and antenna. Slave computers have to use the same base port with the receive address set to the same value as was used for send in the master. (Use defaults if there is only one master on the network.) After exit from this menu the network will be disabled in the main menu. Enable input and output as required ('R' and 'T' in the main menu) and save the settings with 'W'. If you do not save, the old network parameters (par_userint) will be used next time you start Linrad. If they differ from what was stored in this menu an error will be reported. [316]Click this button to enable or disable automatic spur cancellation. [0]