Speech Amp Mic Circuit
Most people prefer to use a high-Z mic like a D104. These mods are designed to work best with a D104 or other hi-Z mic. Anything else can be used once these mods are implemented. Hi-Z mics like to see an input impedance of at least 5 MegOhms or greater. The grid resistor should be anything from 5 to 10 MegOhms, with a 100pf (max.) disk or mica cap bypassing grid to ground, (to keep RF out). Connect a series resistor, anything from 4.7 to 22 kOhm, from the mic input right to grid of tube. The input impedance must be kept very high, while keeping in mind shielding and lead length. Assuming a llow end response of approximately 30 Hz, the cathode resistor should be between 4.7 and 6.8 kOhm. Chamge the cathode bypass to 100mF/50V. The 100mF would normally provide a time constant down to 5 HZ or less. However, good practice requires that tubes heated with AC should use a very large value of Ck or cathode bypass in order to provide a very low impedance path to ground for hum in cathode circuit, (due to heater/cathode leakage).
Change out the 510pf caps to 0.01, 0.02 or 0.05mf caps. What determines the value of coupling cap is the value of the resistor it looking into (the ratio of the two values). For example, in the DX100 second stage the grid leak resistor is a 500 kOhm potentiometer. Five hundred kOhms (500k) and 0.01 uF is good ratio for low end response. Change the cathode ressitor and bypass capacitor to 2.7 - 4.7 kOhm and 100mF/50V respectively.
A negative feedback loop will reduce distortion and further flatten the frequency response. The loop will connect the secondary of the mod transformer to the cathode of second half of 12AX7A. Remove existing 2mF cap from the 12AX7A cathode. [Alternate method (higher gain): Lift ground end of 4.7 kOhms/2mF cathode network, tie together and run to ground through 100-500W resistor.] Increase cathode bypass cap to 50-100mF/50V. From the 500 Ohm tap of modulation transformer, add a 1-2 MegOhm/2W resistor to cathode of 2nd 12AX7A, with a small (0.002uF) across cathode resistor to keep RF components out. Experiment.
B+ Decoupling Caps
Change out the 0.1mF to ground. It is too small. Change to 10mF/450V (preferred) or at least 1mF/600 mylar or film type (quite acceptable).
12BY7A Driver Stage
Negative feedback around driver stage would extend frequency response, and reduce distortion, but gain goes down. You have to determine how much gain you can stand to lose. There is a 100 kOhm resistor connected to the plate (pin 1) of the 12AX7A. Add a 150 kOhm/1W resistor tied from the plate of 12AX7A to the plate of 12BY7A. Change the cathode bypass cap to 50mF/50V. Change the cathode bias resistor to 680 Ohm/5W. If you solid state the LV supply, this value should be changed to 750 - 820 Ohm/5W, and at least 100mF/50V bypass in both cases.
Terminate the secondary into a load, anything from a 10 - 15 kOhm/1W resistor across secondary.
Remove the 1 kOhm resistors in series with grids of 1625's. In normal operation of DX100, the modulator tubes are being tickled into grid current. However, there is no form of bypassing from the center tap of driver transformer, and as a result, the rectified grid current gets kicked up and down on the bias divider string (feeding driver, modulators and final). The normal result is that as you modulate fully, the bias for the driver, modulators and finals gets forced more negative (more towards cut-off), causing grid drive to drop towards zero and the RF power to decrease dramatically with modulation. The center tap (point "J") of driver transformer must be clamped with a 33V/5W Zener diode, bypassed with a 250 - 1000mF/50V electrolytic.
With 250 Volts on the screens (highly recommended), bias voltage should be around 22 Volts. Tetrodes run cleaner with lower screen voltages with regards to intermod products. When the HV supply is solid state and plate voltages are up around the 800 volts plus range, the typical modulator idling current will be around 70 mA, but not more. Remove the 0.02mF/ 1.6KV cap across secondary of mod transformer.
The Turbo-Mod reconnects the modulation transformer as an auto-transforme. Rather than having separate primary and secondary windings, you take the 500 Ohm tap on the secondary and connect it to one of the modulator plates, and the bottom end of the secondary that went to B+ is disconnected. This provides a much lower impedance for the modulator tubes to look into, thus providing a better match. They will deliver more power with less distortion.
RF Final Screen Supply
The screen voltage is derived from a divider in high voltage power supply. This is a poor method. Under normal operating conditions, screen current will cause that screen voltage to sag. In the original circuit, there is a large, 30KW, center tapped resistor, which establishes a half voltage point for the two electrolytics in the power supply, but under normal operation there is an imbalance of voltage, and the electrolytic from the top end of the hi-voltage supply to the center tapped resistor will wind up with over 450 volts across it, boom! - the electrolytic will break down! The screen voltage actually needs to be lowered. The screen voltage should be taken directly from the low voltage supply, through a relay. This is done when the PTT circuit is added. For example, a 4-pole relay is used to key up plate voltage, exciter B+, and modulator screens, as well as audio driver B+.
Final RF Amplifier
The final amp in the DX100 is not operating fully into class-C. Positive peak modulation headroom is not quite sufficient, efficiency is not high enough and modulation linearity is not adequate. The original design was probably used to easy drive requirements. The 20 kOhm screen dropping resistor not high enough. Change it to 30 kOhms. The easiest way is to add an additional 10kOhm/10W resistor in screen circuit.
Change the original 2.2 kOhm resistor (in bias string) to 6.8 kOhm/2W or 8.2 kOhm/2W.
Solid state the HV and LV supplies. Change LV supply to choke input filter, with original 20mF cap now swung over to other side of filter choke, in parallel with 40mF (to make 60mF total). This should provide lower screen voltage to modulators and lower voltage to VFO tube. Ideal screen voltage for the 1625 modulators is between 200 and 250 V. Around 250 volts for screens will work very well. Transmitter will be happier this way. In the HV supply, the 5R4GY tubes are extremely lossy devices, with voltage drop going from 60 to 100 volts. In a choke input supply, when power is turned off, the collapsing fields from choke will cause transients which may destroy solid state rectifiers.
With any choke input supply, rule of thumb is that the PIV rating of the rectifiers should be 8 to ten times greater than DC output voltage of supply. So for DX100 you need 3-4 kV PIV per leg for LV supply and 7 kV PIV per leg for HV supply. "Microwave Oven" diodes are suitable. Put a capacitor and resistor in series across filter chokes. Use a 0.01mF/1kV or 0.02mF/1kV capacotir and a 500 Ohm/2W series resistor for LV choke. Use a 0.02mF/3kV in series with 200-300 Ohm/2W resistor across HV choke to soak up transients. Put a metal oxide varistor (MOV) across primary of plate transformer . This should protect the solid state supply.