So anyway, what ya need to do……
The 12AT7 (V12) is now a cascaded amplifier. Notice the microphonium connector is floating from the front panel. The ground side of the jack is bypassed to ground with a 0.1 uF, 5 VDC disc ceramic cap. This is done to avoid ground loops that will occur between the modulator mainframe and the speech amp. It is useful to have a microphone input for a „just plug it in and playš capability. The usual D-104 or other high impedance type crystal or ceramic microphone offers a good compromise between broadcast quality and good talk power with the component values used.
There is no gain control pot used between the first and second stages. If the gain is too high and overload occurs, a 12AU7 may substituted for the 12AT7 or pot could be incorporated into the circuit. An additional BNC connector is used for a microphonium level input. This offers the use of a coax cable to the front panel and a good positive electrical connection to the input circuit. An RCA type connector may be used if this is the route you choose to go.
The audio bandpass filters and clipper stage (V14) have been removed. In earlier experiments, the 6C4 driver stage was tested with a larger cathode bypass cap (C21) and a larger negative feedback coupling cap (C40 changed from 2700 pF to 0.001 uF). This gave good low end response to about 80 Hz. Then the low end response rolled off as evidenced by a decrease in modulator plate current.
It was noticed, when a frequency response curve was performed using 400 Hz as a reference point, that the response at 3 kHz was down 5 to 6 dB! As the audio frequency was increased, the high end would continue to drop markedly while the modulator tubes continued to pull more and more current, until about 12 kHz. (This is a sign that the frequency response of the modulation transformer is okay but there is excessive amount of shunt capacity somewhere in the circuit, Ed.,HUZ). The problem is in the RF deck. We will get to that later.
The 6C4 driver stage was changed to a triode connected 6AQ5 (low mu) and the driver transformer was shunt fed. (This is also an excellent mod for Viking I‚s and II‚s, Ed., HUZ). A 5k, 25 watt plate load resistor and a 4 uF oil filled cap were installed under the chassis.
Notice that there is a 2 watt resistor across the secondary of T6, the driver transformer. In any situation where the tube being driven are operated predominately on Class AB1, it is very important to terminate the secondary of the driver transformer with a resistor close to its design impedance. This termination helps to reduce distortion caused by the transition from AB1 to AB2 (grid current on voice peaks). A type of clipping distortion can result. (The effect is noticeable on many stock T-368s when the modulation exceeds about 90 percent, Ed., HUZ). Also, the load resistor serves to flatten out the frequency response and rolls off the extreme highs. This latter is important when using negative feedback. If enough phase shift occurs is very high frequencies, ultrasonic oscillation will occur. (The effect on the life of the modulator is left as an exercise for the reader, Ed., HUZ).
With the new driver stage, the low end response appeared to go down to 30 Hz. Not bad for military iron!
Now, back to the beginning again. We have dual inpoots, 1) microphonium level, as previously described, and 2) line level. A 600 Ohm to grid transformer is used for dealing with balanced lines and good RF isolation. However, a typical medium impedance line level output, like that from your usual semi-pro and home entertainment equipment, can be used by going directly to a 25 k line inpoot pot. The two gain control pots (mic and line) are isolated from one another by a 47 k resistor. This makes the gain setting of each independent from the other.
Higher B+ was needed for proper operation of the speech amp. R12, a 10k wirewound resistor that is in series from the low voltage B+ to the audio section, has been jumpered out. Higher value dropping resistors are used for decoupling the speech amp. Further, a 40 – 80 uF electrolytic and a 470 Ohm resistor are used for decoupling.
Negative feedback is now introduced around the entire modulator in the form of five 1 MegOhm, 2 watt resistors from one of the modulator plates (right tube when view from the rear of the transmitter).
It is imperative that these mods are performed or else the high end response will always suffer. In the stock unit, there is about 6000 pF of capacity to ground, not to mention the two vacuum variable for tuning and loading. The circuit was converted to shunt fed by installing a single layer type RF choke, a B&W 800. A National R175 will also work, but may compromise performance on 160 meters due to lack of inductance.
The plate blocking and bypass caps used were 500 pF, 5kV CRL type 858 transmitting doorknobs. The end of L3 (1 mH) formerly connected to the B+ is now grounded. This puts the plate tank components at ground potential. The original blocking capacitors (C41, C42, C7) were left in place. It is a pain to remove them. C10, a 200 pF mica (bypass) was removed.
In the T-368 I worked on, there was an additional pie-wound RF choke in series with the plate current meter (M3). It was a convenient point to tie a length of high voltage wire and connect to the final tube where the other components are mounted (RF choke and 2 caps).
By removing the DC from the tank circuit, a possible source or arcing is eliminated. Further, the total shunt capacity is now about 1000 pF compared to 6000 pF in the original circuit. (This improved to the frequency audio response drastically, Ed., HUZ).
These changes allowed greater plate-to-plate swing of the modulator and therefore better us of the impedance match of the stock mod transformer. The transmitter at this point will do a bit better than 100% modulation in the positive direction. An improvement can be made by substituting 4-250s or 4-400s for the stock 4-125 modulators. The stock filament transformer will weather the storm but it will run HOT. A larger transformer should ultimately be used.
Air MUST be put on the tubes. A 4 inch muffin fan can be JS mounted to blow air between the tubes. Better yet, metal work can be installed to enclose the bottom portion of the chassis that contains the modulator tubes, thus pressurizing that portion of the chassis and blowing air from the bottom of the socket and over the tubes.
With the stock iron, there was a noticeable improvement in the low end response. Further improvement can only be obtained by installing a different modulation transformer. Space is limited. However, there is one type that appears to be made to order, the famous RCA 5500-to 5500 Ohm (1:1 ratio) open frame broadcast type. There is JUST enough space to install it. However, two electrolytic caps (C17, C18) that were part of the carbonium microphonuim voltage circuit must be removed. R19 is also jumpered out of the circuit. The new mod transformer will drop right on the chassis like it was meant to be.
Now 4-250‚s or 4-400‚s are a must. The DC current rating of the RCA transformer is 200 mA. Load the PA to only 200mA. DO NOT load the PA to the book value. The power outpoot is a little low but no one will notice. The transmitter will now easily modulate 150% positive. Al screen VR tubes are used (600 volts). The RCA iron has good, clean low end response and the high end response is as good as the stock iron.
There are possibly better alternatives for the speech amp mod shown here, but the main limiting factor is the mod iron. There is no room for anything larger. Also, 3-500Zs could be used for modulators with the proper audio driver circuitry installed……………