Firmware versions D6 and D6a.
Firmware version D604 March 2006
This version adds some crafty new features - 'Castle' mode, which is like DFCW but faster, a GPS dot-clock option, power level control for Cartesian-control rigs, and a GPS synchronised mode. The description and mods described below relate ONLY to this code version. The GPS dot clock idea will work with older firmware, but all other features require the Version D6 firmware.
Fit a 3-bit R-2R network to PD2, PD3 and PD4, using 1k and 2k resistors. The MSB is PD4, which has 2k to the new POWER output. The POWER output has a 1k resistor to a 2k resistor to PD3; the junction of these resistors has a 1k resistor to a 2k resistor to PD2; the junction of these resistors has a 2k resistor to ground. 5% resistors will do. Wire the POWER output to a new BNC connector on the front panel. The output is 0 - 5V, or 0 - 2.5V into 1k load, in eight linear steps.
Add an RCA connector on the rear panel, labelled GPS 1PPS. Add a pair of 2-pin headers on the board. Connect one pin of each to PD5 (was connected to DOT). Connect the new socket to the other end of one header and reconnect U4 pin 4 (DOT) to the other one. Fit a 2-pin jumper to one of the headers. Obviously if it connects U2 to DOT, you have the status quo - 32Hz dot clock. If you connect the other header, you have a 1Hz GPS triggered dot clock. The beacon script timing values will need to be adjusted accordingly. Set K0000 for 1 sec dots, K0001 for 2 sec, etc. In GPS mode you can't realistically use modes faster than QRSS3, and certainly can't send real-time Morse. For JASON mode, use K0009 (10 sec). The mode isn't fussy. If the GPS unit is external to the Exciter, it would be advisable to include a 1k resistor in series with the new RCA socket to provide a measure of ESD protection.
Mode M2, which was FSK mode, has been replaced with CASTLE mode. In this mode, the dots are shifted low compared with the set frequency, and the dashes high; this might seem the same as DFSK (DFCW), but each successive dot or dash in CASTLE mode is shifted further in the same direction. Thus there is no need for gaps between successive identical elements (e.g. two dots), which makes the mode faster. Indeed, there's actually no need for a gap between letters, as each letter starts near the nominal frequency, but this version does include letter gaps so reading text is easier. The shift is set using the A command. Use about 1/4 of the shift you might use for DFSK (DFCW) mode. The new mode is about 20% faster.
Because of the extra code added, there is no longer room for the message provided by the H HELP command. The command is still recognised, and the processor reset, but no message originates. Version D6b does at least tell you the Version number as a prompt when it resets. Keep the Operating Guide beside you!
GPS Dot Clock
If you select the GPS 1PPS input as the DOT clock (and have a GPS with a good fix connected), you will generate beacon messages that are GPS timed. Each element will start and finish on an exact second event (within a few microseconds). Note that the carrier is not GPS synchronised, in order to keep the keying sidebands to a minimum.
As mentioned above, use timing values Knnnn 1/32 of those used with the 32Hz dot clock generated by U2. By laboriously counting elements, you can set a message that is an exact length (e.g 10 minutes) and so will repeat an exact number of times per hour.
Some LF rigs, notably the Southern Avionics SC1000, have the ability to change power level on the fly by changing the control voltage to the power supply/modulator. The rig thus requires a Phase and Envelope (P&E) drive method. The conventional RF output of the LF Exciter drives the PHASE input, and the new 8-level POWER output drives the power supply. The power is set with the PM command. Set P7 and adjust the transmitter for full output power: settings P6, P5, etc will give lower power. The P command can be used in scripts, so you can send parts of the message at different power levels. For example send 'ZL1BPU 7 6 5 4 3 2 1 0' with the numbers at the appropriate power levels. P0 is the default value, and is zero power.
Typical power levels are (7) 400W (6) 300W (5) 200W (4) 120W (3) 80W (2) 25W (1) 5W. The actual levels depend on power supply linearity and the maximum power setting.
The new P command works so that whenever the transmitter is keyed on, the POWER output is set to the requested power, and when the transmitter is keyed off, the POWER output is set to P0. It is done as the front panel TX LED is keyed. In this way transmitters (such as the SC1000) that don't like to have power applied without drive can be keyed safely.
You might like to add a low pass filter (slug the POWER output with 10uF) in order to reduce the risetime and thus the clicks. This helps at key-down, but since drive is removed instantaneously at key-up, this edge will still be abrupt.
Coherent Carrier Mode
A new mode, M7, has been added. It does not use the script, and can't be included in the script. It appears to operate the same as M0, except that it uses the dot timer and its Knnnn setting to resynchronize the carrier to GPS. Normally (as in M0), the DDS can be commanded to generate a carrier that is within 0.04Hz of an exact 1Hz step frequency. On most frequencies, you can get close to an exact Hz, but never exactly on it, because the DDS steps are a funny value (typically 0.0847710503.... Hz).
By careful choice of operating frequency and synchronization rate, in mode M7 the carrier can be made to operate at any exact Hz step. The carrier is first adjusted as close as possible to the desired frequency in M0 (choose the closest frequency, it can be high or low), and then set K0000 and M7. Then increase K slowly to find the best update rate. Listen for glitches as the carrier is synchronised and try to minimize them. This is best done with a Spectrogram - ARGO in 10 sec mode, for example. K0009 (10 sec) should work, implying the frequency was within 0.1Hz in the first place.
This mode involves phase synchronization: it is not a phase-locking technique, and so is simple, but clean and fast. The carrier phase is reset to a constant value in the dot clock interrupt. If the frequency is set accurately and an appropriate K value chosen, there will be minimal phase shift at the interrupt, and thus minimal sideband generation. Note that some frequencies will give better results than others, depending on how close the nearest DDS step is to the chosen frequency.
Obviously this feature MUST be used with a GPS receiver with accurate 1PPS output. It does not work with the dot clock generated by U2!
Another interesting point is that the technique can also be used to lock to fractional Hz values. A simple example is to use a multiple of 2 for K, remembering that values are offset by one (e.g. use K0003, K0007 etc), and lock to 0.5Hz steps. The use of K values greater than K0007 will require an exceptionally stable reference, such as a good TCXO. Variations in the reference may result in varying levels of sync-rate sidebands, or even lock-rate steps in frequency. The author uses K0009 with a simple 12.8MHz TCXO with complete reliability.
At this stage there is no ID or beacon capability in Coherent Carrier mode. If you externally key the POWER output (key in series with the POWER output) you can manually key the transmitter and still stay in phase.
There is a new version of the 'Make Beacon' program (MAKEBCN2.EXE). The major addition to the program is the addition of a 'literal' capability so you can add binary commands and binary data into the scripts. This will allow easy manual generation of scripts which include JASON, MT-Hell (even simple pictures) or ASK Hell. Other modes, DFSK (DFCW), QRSS and CASTLE are supported with full compilation capability.
The literal mode is entered with the script command '$L' and exited with the command '$l' (lower case L). Anything between these commands will be included in the beacon message verbatum; the data won't be interpreted as text or commands.
The message length checking has been removed in this version of MAKEBCN2. This feature was no longer reliable when literal message segments were used, since some take up space without generating data. You'll have to manually check by trying the message out whether it is too long or not.
Note that MAKEBCN2 does not at this stage support the $P command in the script - you have to code power settings using the 'literal' capability, e.g. '$LFB06$l' sets power level 6.
Here's an example of a script with all these features:The message includes CASTLE mode and power level shifting. It's not intended for a GPS dot clock!$M1$LFB07$l $K000D DE ZL1BPU BCN $M2$K00C0 BPU 7$LFB06$l6$LFB05$l5$LFB04$l4$LFB03$l3 ~
32bit Version10 March 2006
Version D6b uses a 32-bit synthesizer, which allows 0.3mHz (that's milliHz) resolution. In other words the resolution is typically considerably better (1 e-9) than the reference accuracy (2 e-6). The main loop operates using 10 clock cycles rather than 9, and so the maximum operating frequency is reduced by about 10%. Use 10 rather than 9 when making calculations! The step resolution can be calculated from:
Resolution = Crystal Frequency / (10 x 232)
No change has been made to the offset and sweep features - the step sizes and commands remain the same. All that happens is that the F commands become longer (8 characters) and the resolution is increased.
Now, with GPS locking (to a 1Hz or 1pps signal), you can operate on just about any frequency which has a simple relationship with 1Hz, say 1/8th Hz or 1/10th Hz steps. However, it's often desirable to operate at an integer Hz frequency. The Exciter with this firmware will generate numerous 'round figure' integer frequencies exactly (within the reference accuracy). For example, with a 12.8MHz reference, the following 'round figure' frequencies are possible:You can also lock at 0.2Hz steps if you know how (K0004), with exact frequencies every 125Hz. These supposedly exact frequencies have lower unwanted products (spurs) than others when locked. Remember that their actual accuracy depends on the calibration of the reference clock.135.000 138.750 180.000 135.625 139.375 180.625 136.250 140.000 181.250 136.875 140.625 181.875 137.500 141.250 182.500 138.125 141.875 183.125 kHz
The exact frequencies with GPS locking and K0000 are spaced every 625Hz. Much the same applies to some other crystal frequencies: for example, with a 16MHz crystal, the spacing of exact frequencies is 781.25Hz. With a 10.24MHz crystal, exact 1kHz steps will be generated, but only up to 250kHz. Remember that of course millions of other frequencies can be generated, as the frequency resolution is less than 3mHz, but none of these are integer or simple fraction accurate. For most applications this is irrelevant, and besides the resolution (about 1 part in 109) is much finer than the oscillator stability with all but the most exceptional oscillators.
Version D6b has been enhanced for GPS-coherent precision applications. It adds the following new modes:
- Mode 7 This GPS-synchronized mode now has controllable phase. Phase can be set from 0° to 360° in steps of 360/256° using the Ann phase command, which in this mode replaces the Ann Frequency Offset command. The actual phase generated for 0° depends on the operating frequency, due to the fixed GPS interrupt delay. Phase is reset to the commanded value every Knnnn+1 seconds. Values of K up to 20 may be useful, depending on the quality of reference used. K=0 causes reset every second.
With the internal 32Hz dot clock selected, this mode is meaningless.
- Mode 8 is a PSK phase reversal mode. It transmits fixed GPS-sync PSK phase reversals that makes the signal highly identifiable. No data is transmitted. The Knnnn command is used to set the interval between phase reversals. The reversals are made GPS synchronously at an integer number of seconds, 1 - 65525, so in theory 65,000 or so different signals could be generated. Phase accuracy is sufficiently good that the nominal carrier is well suppressed, and two sidebands spaced either side by the keying rate predominate(e.g. ±1Hz for K=0).
With the internal 32Hz dot clock selected, the reversals are up to 32Hz rate, and of course not GPS synchronous.
Mode 7 is useful for 'Click Lock' applications. This highly sensitive technique was developed by Peter G3PLX, and allows an uncalibrated but stable receiver to monitor GPS-synchronous signals and integrate the received energy for long periods (up to hours). The following example shows Click Lock software by Con ZL2AFP indicating the phase performance of the ZL1BPU LF Exciter in M7 mode with K=15, A=0, operating at 200kHz. The integration time was about 1000 seconds, and the signal was so weak as to be inaudible on the receiver and invisible on a spectrogram.
You're not likely to know what you're looking at here! The right-side plot is an integration of the In-phase and Quadrature-Phase products of the received signal over a period of 1000 seconds. That the black line which records signal phase is approximately straight indicates that in the measured period the phase changed very little. The slight scolloping of the trace shows how minor frequency errors (which cause curving of the phase plot) are corrected. Transmissions made with these settings will ALWAYS move in the direction shown on the vectorscope display, although the pattern will rotate as the distance from the transmitting station increases, and will also rotate with changes in the ionosphere. The next image shows the same effect in higher detail (over about 100 seconds).
The appearance of the phase plot is not ideal, as it is limited by the technology involved in micro-based DDS. It does limit the ultimate S/N performance to some extent (the signal bandwidth - a few milliHz - is wider than ideal), but it should be useful to integration times up to several hours. One thing is for sure - the signal is uniquely identifiable!
This technique is so sensitive that the ZL1BPU LF Exciter has been copied and reliably identified barefoot at a range of 500km! The EIRP of the transmission was 80uW!
Command Addendum, Version D6bThere is no separate manual edition for this 32-bit version, so make note of these changes. Maybe make a copy and slip it into your hard-copy manual.
- Ann OFFSET
- Defines the phase offset from a nominal zero phase angle in GPS-synchronized Mode 7. nn is a hexadecimal value representing phase 0° to 360° in steps of 360/256°. nn can be any value from 00 to FF. Ann has no effect on frequency offset in Mode 7 or Mode 8.
- Fxxxxxxxx FREQUENCY
- Now that frequency control is 32-bit, the F command now has a 32-bit (8 hex character) value. Remember to use the 32-bit resolution calculation when defining frequencies, or you'll be out by a factor of 9/10.
- Knnnn KEYING
- Defines the rate (period in seconds) at which GPS phase corrections are made (Mode 7) or GPS phase reversals are made (Mode 8). nnnn is a hexadecimal value representing one less than the number of seconds between phase corrections. The range is 0000 (1 sec) to FFFF (65536 sec), but values larger than 0010 are unlikely to be useful without a very good TCXO or OCXO reference driving the micro.
- Mx MODE
- Mode M2 is now Castle Mode. Keying speed and Offset are set as in DFSK mode, except Offset should typically be about 1/3 to 1/4 of the DFSK value.
Modes M7 and M8 provide GPS-synchronized operation at a user-defined phase angle (M7) or two arbitrary phase angles about 180° apart (M8), when GPS 1pps pulses are used as the dot clock. The rate of phase correction is set by Knnnn, and in Mode 7 the actual transmitted phase is set by Ann. The beacon script is not used in these modes, and they cannot be called from a script.
- Px POWER
- Sets the transmitter power output level. A controlled voltage 0 - 5V in eight steps. Value P0 sets zero power, and P7 sets maximum power. When the transmitter is keyed off, P0 is used. When keyed on, the user set Px value is used. Any value of x from 0 - 7 is valid.