Heising modulation, also known as constant current modulation, was very popular in the early days of AM. It was largely replaced by conventional transformer coupled modulation systems. It is important to remember Heising modulation does NOT use a modulation transformer.  Contrary to popular misunderstanding, a dc current bypass choke used to prevent PA steady-state current from flowing in traditional transformer modulated stages is NOT Heising modulation!

The Globe Scout series is one of the unusual or uncommon systems in the later days of AM  using Heising modulation.

By definition Heising modulation has a 1:1 ratio modulator to PA impedance ratio. Heising modulation is a constant current modulation system, defined by current drawn from the power supply remaining constant as long as the modulator is not distorting the audio.

Here is the Heising circuit I use in my modified Globe Scout 65A.

Using the circuit above as an example, we can see why Heising modulation is difficult to work with. Heising modulation operates in the following manner:

The modulator tube, like any tube, acts like a time-varying resistance. The anode-to-cathode resistance pulls the voltage at point A up and down at an audio rate. The large inductance of L1 allows this to happen. The collapsing or expanding field in L1 actually allows the voltage to go above the B+ rail. (But not as cleanly or far as it would with a transformer and push-pull modulator, where the voltage can be forced up.)

Another way to view this is the modulator and PA tube share current through L1. A basic property of inductors is they try to adjust voltage to maintain constant current. While capacitors try to maintain voltage across terminals with varying current, opposite behaving inductors try to maintain constant current through terminals while allowing voltage to vary!

If the modulator current decreases, the collapsing field in L1 causes and increase in voltage and the PA tube has higher operating voltage. Current that formerly went through the modulator tube now goes through the PA tube, along with the higher voltage.

When the modulator tube draws increased current, the choke tries to oppose the change and generates a counter EMF opposing the supply voltage. In this case PA and modulator voltage is reduced, and current that was in the PA is now diverted to the modulator tube. 

Linearity and Distortion 

The modulator tube and reactor must never be allowed to go into non-linear regions. The resulting harmonic and intermodulation distortion will appear (unfiltered) as modulating voltage and cause splatter.

One test of proper linearity is to measure current from the supply through the modulation choke, L1. If the supply current flowing through L1 current changes at an audio rate, the modulation cannot be linear. This is why Heising modulation is called constant current modulation. Of course we can use conventional and easier-to-apply distortion measurement techniques also, but this is a novel characteristic of true Heising modulation.

There are also limits in voltage. Even if the modulator tube is saturated with positive grid voltage, the modulator tube could never pull voltage at point A lower than positive voltage at point C. As a matter of fact, the lower voltage limit at point A on negative cycles is always greater than the positive voltage available at point C. This limits modulation to less than 100%, unless a dropping resistor is included somewhere between the modulator anode and the PA voltage feed. The resistor can be at any point that limits  anode to cathode voltage in the modulator to a value significantly less than the lowest voltage obtainable at the choke and modulator anode connection junction with full linear positive-going modulator grid voltage.

Since audio rate current flows through the required dropping  resistor (R101), it is imperative the dropping resistor is bypassed with a capacitor having very low impedance compared to the resistance value of the resistor. Without the bypass capacitor the cathode will lift on negative modulation peaks (when the grid of the 6L6 goes positive) and reduce negative modulation peaks. This is a downfall of the original Globe system. They never included a low-reactance cathode bypass capacitor. That probably is a good idea for CW, because the CW waveform might suffer. It is a bad idea for AM, however.

Since modulator and PA anodes are in parallel for audio signals, the anode voltage swing in the modulator and PA track each other. This means the 6146 PA has voltage swing limitations similar to the modulator tube.  Because of the voltage limit established at point C, the 6146 can't possibly reach zero anode voltage. The solution is the addition of R106 and C110. The resistor causes the anode of the 6146 to operate at reduced voltage, and C110 provides a low-impedance path for audio variations. 

For 100% linear modulation the screen must also follow audio modulation of the anode voltage. This is accomplished by operating the screen from the same source feeding the anode. C113 has the effect of increasing audio voltage level applied to the screen. 

Screen resistor R107 may or may not require bypassing with C113. There almost never is a case where full modulator voltage would be applied to a screen. If C113 is required to achieve 100% linear modulation, the screen's audio path impedance can be limited with an additional resistance (not shown) in series with C113.  (Rather than adding a resistance in series with C113, the resistance of  R107 could be split between two series resistors. C113 could then only bypass one resistor. R107 could be a tapped adjustable resistor of the proper value to establish proper screen voltage, with C113 between the slider and the end. 

My Scout

The circuit above is the actual circuit used in my modified Globe Scout 65A. C113 requires a large series resistance in my Scout, about 220K ohms. The resistor value was determined by watching modulation linearity. I use a sine-wave audio generator, comparing the diode-detected audio of the RF output to the generator audio output.  A standard trapezoidal pattern would work also. 

R101 sets the quiescent current of the 6L6 at the maximum allowable dissipation of the tube. I used a 6L6GC, which has higher dissipation than a regular 6L6. I adjust R105 to set the 6L6 at the maximum dissipation limit.

I generally operate the 6146 PA at the same current as the 6L6 while operating AM. This modified system allows 100% negative peaks and nearly 100% positive peaks with voice. PA current is about 55-65 mA, or 30 watts. In class C operation, my Globe Scout runs about 18-20 watts carrier output.  

Notice R109. This resistor is a 2 watt high voltage type resistor that connects to the cathode of the 6J5 triode in the audio amplifier. The bypass capacitor is removed by snipping the white wire at the cathode pin. The 6J5 is a weak link in the Globe audio system. Bias is too low. On positive peaks, the 6J5 grid actually goes into conduction. This clips the signal from the gain control, and limits positive peaks. 

R109 does two things. It biases the 6J5 closer to cutoff, allowing the grid to swing more positive before drawing current and limiting. It also adds negative feedback, reducing non-linearity in the 6L6 modulator tube.

I tried removing R101 and using grid bias to reduce the lower-voltage limit (because point C could be chassis ground), but the actual gain in negative peak linearity was too small to make the mod worthwhile. 

I also added a grid leak resistor, a screen clamp set at 225 volts, and a fixed grid bias source. The fixed bias is set so the Globe draws about 40mA with  HV and no excitation. It is important you do this, because the original 6146 cathode resistor of 450-ohms adds negative bias that decreases modulation level and linearity. If you want to leave the 450-ohm resistor in, I suggest adding an electrolytic capacitor for audio frequency bypass. In that case it would not be harmful, and you could omit C110 and R106. I have not tried this because I converted my Globe to grid block keying and eliminated the 450 ohm resistor long before starting audio modification. 

With no other mods the Globe Scout is flat from 200Hz up to about 3500Hz, where it starts to roll off. By 6000 Hz audio is 6dB down. The Globe Scout audio is comparable to any other good radio (Ranger, etc.) on AM.

This is the original Globe Circuit. (The penciled-in components were from a mod I made in 1963) :

Since creation 2004 9/26