It is assumed that you have already read the linking pages.  If not, go back and read them again.  It would also help to read a full description of the Amplitude Modulator version of the transmitter described here.

The described circuit is the very first version that I built.  It is not really optimized in any way, but was built to test the concept, and the hardware, and the software, and...

Designing the circuit:

Figure 1:
Top:  The prototype PIC-based PSK31 board using a balanced modulator (the very first one.)
Bottom:  The schematic of the prototype exciter.
Click on an image for a larger version

For the following discussion, please refer to the schematic - see the link to the right under the picture.

A few comments about the circuit:

Please note that in order for the circuit to work as drawn, there are a few things to keep in mind about the op amps being used.

• Use a (more expensive) op amp that can swing up to each rail - that is, a "Rail-to-rail" op amp (both input and output!)
  
• Use an op amp that can swing near one rail (i.e. the LM324 can go nearly to the negative rail) and "shift" the bias to favor that region
• Set the gain/offset of of the op amp such that its output stays far enough away from each rail to prevent distortion.
• Or, you could just use a negative supply!
  

Balanced Modulator Method:

Using a balanced modulator is an obvious solution:  It can modulate both amplitude and phase in exactly the way that is required for PSK31:  One need only apply RF and the appropriate baseband modulating signal and the result is a PSK31 signal. This can be done with either a diode modulator (a singly or doubly-balanced modulator) or an IC of similar function, such as the venerable '1496 or even the NE602.  When driving ICs such as these, careful attention must be paid to proper drive levels, biasing, and modulator balance to obtain best results.

This method works well to generate low-level RF signals, but linear amplification is necessary to develop additional RF power.  If, as in the case of a LowFER or MedFER beacon, transmitter efficiency is highly important (keep in mind that Part 15 specifies input power) linear amplifiers would result in 1-2 db lower power than could be obtained from a high efficiency switching-type P.A.  In addition to using it for generating low-level RF, this circuit could be modified to produce an audio signal that could be piped into an SSB transmitter, for example.

A brief description of the circuit:

The method of generating the analog baseband signal via PWM is discussed elsewhere, but suffice it to say that when a "0" is to be sent via PSK31, a phase shift is used to represent that bit.  This means that the baseband output changes state from low to high, or from high to low.  This change doesn't happen instantaneously, but rather it does so in a "spectrally friendly" way:  It's voltage changes as a cosine.  What this means is that if a constant stream of 0's were sent (alternating phases) the baseband signal would be a 31.25 Hz sine wave.  This is what gets applied to the two diodes.

U2 is wired as a unit-gain bridging amplifier.  One section does a mirror-image of what the other does.  In this way, twice the voltage swing is available across the outputs of the two amplifiers than would be available if just one were used and ground (or power-supply) referenced.  Why just 5 volts?  Part of the idea was to allow this to run on 4 AA batteries and be able to tolerate voltage droop.  The regulator was added just as a convenient way to allow it to run from a 12 volt bench supply.

The two (almost) back-to-back diodes form a cheesy balanced modulator.  Ideally, if there is no voltage across the diodes, they are shut off and there is no output.  Even if they were to "leak" a bit, they would leak equally and the resultant leakage would be canceled out.  U3, a 74HC00 functions as a simple buffer/inverter that supplies equal and opposite RF phases to each diode.  When the input of the bridging amplifier (consisting of U2) is operating at one logic level, there one of the diodes is forward biased (allowing it to conduct RF) and the other is reverse biased (blocking RF.)  When the input of the bridging amplifier is in the other state, the roles of the diodes are reversed and the RF being conducted by that diode is of the opposite phase.  In intermediate voltages between the two states, the amount of signal conducted by the diode that is turned on is proportional to the current flowing through it (assuming you don't ask for more signal than is available...)

You'll no doubt notice that there is no balance control for this "balanced modulator."  Since this is a prototype/testbed circuit, I didn't bother adding one.  A balance control really isn't all that important, anyway, if all you are going to send is PSK data because the carrier only passes through zero output on its way to the other phase/amplitude and doesn't really spend any time there.  The side effect of having a somewhat non-balanced modulator is that the RF output of the two phases may be somewhat unequal.  There is also the fact that unless there is a precise 50% duty cycle, there will be unequal power in the two phases anyway.

The output of the balanced modulator is very low (only  a few volts peak-peak,) of high impedance, and rich in harmonics.  If you intend to put this on the air, you will definitely want to put a bandpass filter on the output prior to linear amplification/buffering.  You have been warned!

Originally, with the AM version, the IMD reading was only 18 or so.  This was partly because of the op amp being driven too close to one of its power supply rails, thereby clipping the baseband waveform slightly.  Another reason is that the extremely simple R/C filter and nonlinearities in the PWM conversion caused a bit of distiortion of the waveform.  The table was "tweaked" and some predistortion was added.  The result?  A very clean waveform, rivaling the best of those on the air.

Figure 2:
Waterfall/IMD reading from the PSK31SBW program, version 1.08.  Note the inverted "vee" caused by oscillator phase modulation and propagation delay in the logic gate.

From the "phase-o-scope" display one can see that there is a slight problem with this circuit that does not occur with the as-built AM version:  Oscillator phase modulation.  While we do want a phase modulation to occur, we want it to occur only in the modulator section.  I noticed a tendency for the phase-o-scope on the PSK31SBW program (see picture to the right) to display an added phase shift.  This is manifested by the "peace sign" on the phasor scope.

There are several apparent causes of this:

• Oscillator phase modulation.  This can be a result of slight modulation of the power supply, and/or circulating RF currents.  The original "original" version actually used one half of the 74HC00 as the oscillator the other half as the 180 degree inverter and there was probably 40 degrees of undesired phase shift.  I resorted to using an outboard oscillator/buffer to reduce this effect.
• The design of the dual-phase output itself.  If you look at the schematic, you'll note that there are two phases of RF required to drive the balanced modulator.  These are generated by using one of the 74HC00 gates as an inverter.  That extra inverter stage adds a bit of gate delay (7 nanoseconds or so) which manifests itself as a phase shift.  This amount of delay corresponds to approximately 13 degrees of phase shift at 5 MHz, the approximate test frequency of this circuit (and this would get worse with increasing frequency!)  One way to avoid this problem is to use a different sort of balanced modulator that does not require two RF phases such as the 1496, an NE602, or a diode-ring DBM.  Additionally, using faster logic, or possibly using a different logic configuration (such as using XOR gates wired appropriately or using different outputs of a divide-by counter) can greatly reduce this differential delay.

Alternative methods for PSK31 generation using a balanced modulator on the Small Wonder Labs PSK-20:

Back in 2003, Ralph, W0RPK, approached me about the modification of the now-discontinued "PSK" series of radios by Small Wonder Labs.  His need was to generate fairly high-power (several watts) of clean PSK31 signal for a beacon transmitter - but to not use a full-blown computer for this.  Since he already had the PSK20 radio on hand, that was a natural place to start.

As you might guess, these can apply to about any radio in the Small Wonder Labs PSK series - not to mention applying to about any other SSB-type radio that you might have that uses a balanced modulator!  It should go without mentioning that in this application, transmissions were "one way" and the receiver portion was unused.

Here - in a nutshell - are the modifications that were made to the PSK20 to use the above citrcuit (lower part of Figure 1) to generate PSK31 signals.

• DC-isolate U6, the mixer.  "Unground" U6 the TUF-1 diode mixer.  Make sure that all ground paths are removed.
• Estabilish a new "RF" ground.  Put a 0.1uF capacitor between U6 (the mixer, pin 3) and ground
• Adding a series capacitor.  Insert a 0.01uF between pin-1 and the junction of R41/R42 of the PSK-20.
• Apply modulation.  Remove (or at least disconnect one end) of C24 of the PSK20.
• Connect the modulator.  Connect the newly-isolated "ground" of the PSK-20's U6 (mixer) to "R10" of the diagram in Figure 3, and also connect the "mixer" side the PSK-20's C24 junction (the former audio source) to the "R8" connection in Figure 3.
• Offset the PSK-20's carrier oscillator.  Previously, the "audio" was intended to go from around 500 Hz to 2500 Hz, but with our "baseband" modulation, we need to be near "DC" - that is, the carrier oscillator needs to be tuned within the passband of the PSK-20's crystal filter.  Fortunately, there's a trimmer capacitor that does this!
  

Now, it's time to adjust things. Refer to the diagram in Figure 3:

• Set R5 for minimum gain (e.g. lowest resistance.)
• Set R11 for minimum "drive"  (wiper toward U2D.)
• Set R9 for 3-5 volts on the output of R9.  The exact voltage isn't critical.  If you like, you could replace R9 with a pair of equal-value fixed resistors in the 10k-100k range.
• Connect the PSK transceiver to a dummy load, turn it on and key it up.
  
• Power up the PIC circuit.  Using another receiver to monitor and lightly-coupling it to the PSK transmitter to prevent the receiver from overloading, tune in the signal from the PSK transceiver and look at it using your favorite PSK-31 receiving program.
• Note:  Make sure that the noise blanker on this monitoring receiver is turned OFF and also set the AGC to FAST to get the "truest" possible reading of the PSK-31's signal quality.
• It will also help to turn off your radio's preamp (or, in the case of some radios, turn on the "IPO" mode) as well as turning the attenutator ON.  The idea here is to get a weak enough signal to that receiver to prevent it from being overloaded and giving a false "bad" reading - not to mention detect a "true" bad signal from the PSK transceiver!
  
• Null out the carrier.   Adjust R3, the OFFSET control, for minimum RF output power.  You will have some weak PSK-31 modulation, but ignore that:  Look for the lowest overall reading on the wattmeter and/or the lowest S-meter reading on your  receiver.
• Slowly increase the drive level.  Increase the gain using R5, the "gain" control and noting the power output.  The power output of the PSK transceiver should increase, but at some point, you'll notice that the RF wattmeter will note the maximum rated power of the PSK transceiver and that signal quality on your monitoring receiver will start to decrease.  Pay particular attention to the "IMD" reading!
• Increase power to the point that the IMD reading starts to decrease and you not the signal quality decreasing as evidenced by "splatter" on the waterfall.  Now, back off the power again until the signal "cleans up" - and then, noting the wattmeter level, decrease the power another 25% for good measure!
• Re-check the carrier oscillator tuning.  With the PSK exciter running normally, readjust the 9 MHz carrier oscillator frequency of the PSK transceiver for maximum RF power output, setting the PSK generator's R5 "gain" control to reduce power to the level obtained in the previous step.

Figure 3:
  A version of the PIC-based modulator showing the connections to the modified PSK-20 transceiver - see text.
Click on the image for a larger version.

"Where's the software for this?":

The source code for this is written in both C and assembly.  Go to the About the Source Code page for a description and a link to the actual source code.  Note that both versions of the transmitter (the Amplitude Modulator and Balanced Modulator) use the same software.  The "mode" is selected by the state of PB4 on the PIC:  If it is open (high) then the AM mode is selected.  If grounded, the balanced modulator mode is selected.
References:

For additional links/information on PSK31, MedFER, and LowFER operation, refer to those at the bottom of the PSK MedFER page.
 

Return to the PSK MedFER page

Any comments or questions?  Send an email!