|
Figure 1: As discussed on the previous page, the modulation process of combining two phase shifted waveforms gives rise to a sinusoidal function. In this oscillogram, known as a trapezoid, we have the baseband "raw" audio applied to the x-axis (horizontal)
in the form of a 1khz, sorry, 1kc/s sine wave. The Y (vertical) deflection is the modulated RF carrier. Horizontal deflection to the right is caused by the negative peak in the audio, thus giving rise to a extinguished carrier. Deflection to the left is caused by the positive peak of the audio, giving maximum RF enevlope.
Zero audio modulation of the carrier corresponds to just to the right of the Y-axis (take my word for it). Note the curve of the modulation - a linear modulation process would result in a triangular pattern with straight edges.
In technical terms..... The unmodulated carrier is 4 "squares" high on the scope (just to the right of the Y axis, remember?). Maximum audio modulation in a negative sense (about 2.8 horizontal squares to the right) results in no carrier. Maximum audio modulation in a positive sense (about 2.8 horizontal squares to the left)
results in a maximum carrier of about 5.5 verticals. Thus with 100% input, we have only managed to increase the carrier amplitude by 1.5 squares out of 4 - about a 1.37 times increase. On the previous page we determined that the +ve peak should actually be 1.85 times the carrier (You did pass your high school maths exam, I assume?) So why do we have squashed peaks here?
Well there may be several reasons for this. Firstly, the wave displayed was obtained by resistively combining the two modulator outputs - this is not how the system was intended to operate. The phase shift between the two waves gives rise to complex currents, for which compensation must be made. One output will see the drive of the other stage either 90 degrees before or after its own output, hence this appears as an inductive or capacitive load.
Also, at present I have no way of determining if the system is set exactly for 135 degrees of separation, and finally, it could be something to do with the fact I set the modulator up to produce a nice display on the scope to photograph!
(As an intersting aside, if you extend/project a diagonal line from the zero point, following the initial angle of the curve towards the left, the point at which this projection passes the limit of audio modulation corresponds to a height of 8 squares - eg. twice the unmodulated state, just what you would expect in a perfect world)
|
Figure 2: Squashed peaks anyone? This figure shows a good old AM modulation envelope of a 1Khz tone. Due to the fact that I am directly coupling the two phase outputs of the modulator, without any form of harmonic filtering, the peak negative condition (zero carrier) is a little rough, due to 2nd and 3rd RF harmonics, but I digress.
Take a good look, and prepare for more maths...... (I did warn you this was not for the faint hearted!) Look at the null near the centre of the graph, roughly on the Y axis. Now look up by one square - the width of the waveform at this height appears as fractionally under one square wide. Now look at an adjacent crest. Look down one square,
and the width of the waveform is just over 1.5 squares wide. This is a very good example of squashed peaks. Indeed the peak crests should be about another square higher. For refernce purposes, the Y shift and Y gain in this oscillogram are set the same as for the Trapezoid above. The unmodulated carrier being 4 squares high.
Have you spotted an as yet unexplained feature of the modulation? Look carefully at the amplitude of the positive peak and of the negative peak from the zero axis. The negative half cycle is approximately 0.5 squares less than the positive. As yet I have no explanation for this assymetrical modulation. It may be due to amplitudes of the two phased outputs being unequal (although I did make them as equal as possible), or perhaps is due to RF harmonics, or maybe it's just my 'scope getting a little tired. Any learned suggestions will be most welcome! |
 Figure 3: Tired of the theory? Lets put it into practice. Here we have some real audio (pun intended). This was an audio feed from the weekend satellite programmes on Sunday 30th January 2000. During his programme, Bob "Buzby" Lawrence played a Doobie Brothers song. I will award a prize (as yet undecided, but probably a couple of dud valves from the TX -
real collectors items, be the envy of all your friends!) to anyone who can identify the song from this few milliseconds of envelope. And just to make sure you are not guessing at the title of Doobie songs, you must tell me at which point in the song this was taken (to within 100 milliseconds or so).
|
 Figure 4: We previously discussed the broad audio modulation capabilities of the ampliphase technique. Waveforms like this are normally found only in text books, and very rarely seen in a transmitter. Given that the carrier frequency is 819Khz, you should be able to calculate that the audio modulation is somewhere in the region of 35Khz. This shot was taken from the
output of the modulator, whether or not the whole TX could manage this remains to be seen. Very few transmitters could manage this, older plate modulated ones certainly couldn't, and only the screen-grid and modern wideband, digitally modulated ones could try. The modulator was un-modified when this shot was taken, ie, no filters were removed. This is the natural bandwidth of the system. And it's true what the books said abaout sidebands. Using a radio receiver you can tune
through the carrier and two side-frequencies 35Khz either way along the dial.
|
As the heart of the ampliphase principle relies on phase modulation, Lissajou figures taken between the two channels are essential in the correct settting up and diagnosis of the transmitter.
 Figure 5: Here we see a Lissajou figure showing a fixed 90 degree phase difference between the X and Y axis. In an ideal world this would be a perfect circle, but there are various harmonics present in both drives, plus an amount of phase shift inherent in the 'scope. On this scope in particular, the X scanning is performed by magnetic scan coils, and the Y by conventional electrostatic plates. At a carrier frequency of 819Khz, differences in the performance of these systems are starting to become apparent. A difference of 90 degrees corresponds to the instant of 100 % +ve modulation peak of the ampliphase system.
|
 Figure 6: This is a phase difference of 135 degrees. This corresponds to the "quiescent" setting of the carrier for normal modulation. The sine of 135 degrees (0.707) is equal to the height at which the waveform crosses the Y axis (1.5cm) divided by the maximum Y height (just over 2 cm)
|
 Figure 7: This is a display of 180 degrees difference between the two channels. Note the slight distortion between the two waveforms due to the aforementioned harmonics, etc, which prevents complete cancellation. A phase difference of 180 degrees corresponds with the instant of 100% -ve modulation. If we designate this trace to be a slope of -45 degrees,
a phase difference of 0 degrees, would give a trace slope of +45 degrees. In practice, it is not possible to adjust the modulator to give a quiescent output of 0 degrees.
|
 Figure 8 : This is the 135 degree carrier of figure 6, being modulated to approx 20% by a 1Khz tone. Note that on -ve peaks the width of the central "hole" is narrowing (towards the 180 degrees of fig 7) whilst on +ve peaks, the total outside width of the display is widening towards the circle of Fig 5. In reality, the scope display is moving between the "narrow" trace and the "wide" trace 1000 times per second.
|
 Figure 9: This is the 135 degree carrier of figure 6, being modulated to approx 80% by a 1Khz tone. Note that the minimum width in the centre is now very close to 180 degrees, and the outside width is very close to 90 degrees.
|
 Figure 10 : Hey Presto! We have 100% modulation. The negative peaks now completely close the "hole" whilst the positive peaks now produce a 90 degree circle. If only life itself could be so simple and perfect......
|
 Figure 11: This is a seriously overmodulated 135 degree carrier. The 180 degree hole is now completely closed, and has started to expand (we have driven the modulator past 180 degrees, and the output is starting to reform as it no longer cancels out). The 90 degree circle is now stretched towards a 0 degree slope, as the positive peak is somewhere between 0 and 90 degrees.
When driven from an assyemetric audio source, ampliphases were rated by RCA for producing at least 125% positive modulation peaks at full power. The audio processing must be carefully aligned to ensure that negative peaks never exceed 100%. |
 Figure 12 : If the quiescent carrier is set for 90 degrees difference (as per fig 5) then negative modulation gives 180 degrees and positive modulation gives 0 degrees. Although this may sound advantageous, in practice this gives serious compression to the positive modulation, due to the co-sinusoidal addition of the two waves in the combiner. See here for an explantion.
|