|
Earlier posts on the AM Window and discussions by me and
others as applied to DX-60 carrier control modulation scheme led me to re-think
the system. After a few cut and try efforts, a bit of pencil and paper work,
I proceeded to change several components in my trusty Heathkit DX-60. This
process closely follows the work previously done by KS3K and WC3K, but refines
several values to enhance performance.
Explanation of Mods
Referencing
the cathode (pin 9) and grid (pin2) of the 6DE7 modulator to the -140
volt internal bias supply, through a voltage divider and zener,
provided the ability to modulate 100% positive and 100% negative.
The zener diode arrangement was found to be necessary and a nice
improvement to earlier efforts. It was found that during
modulation, the bias voltage did vary, due to increased current demands
within the power transformer, and thus distortion was higher. The
selection of resistor/voltage values established the no signal resting
carrier about 20 to 30 watts output. This target was chosen as
most linear amps today exhibit some 10 to 12 dB of gain and thus 25
watts input should net some 250 watts output. This is correct for
a 1500 watt PEP amplifier under AM conditions.
Changing
the 6146A screen bypass capacitor to a smaller value (0.005uF to a
0.001uF) and eliminating other bypass capacitors in the audio section
enhanced frequency response. I found that by increasing coupling
capacitors to larger values over the original design values
improved low frequency response.
By
adding an RC feedback network from the cathode (pin 9) of the
modulator, 6DE7, to the input stage (pin 3) of the 12AX7 reduced
distortion. Raising the voltage amp cathode (pin 8) of the 12AX7
above ground, and on the 6DE7 (pin 8) adding a bypass capacitor further
reduced distortion.
The
loading resistor (15k ohms) on the 6DE7 (pin 9) connected to the -140V
bias was the critical element in getting the high frequency modulated
distortion down. This value can be reduced further to about 12k
but I would not suggest going lower to 10k, as this will exceed the
power dissipation of the 6DE7. Elimination of the RC
network (R32 & C35) between the 6DE7 cathode and the mode switch
enhanced the low frequency response and improved modulation linearity.
The Results
Frequency response:
From mike input to about 100% modulation is +/- 2 dB from 35 Hz
to 16 KHz*. I observed that the 100% positive modulation power
level is the same peak level as the CW tune power level or about 65
watts output. Also, the positive and negative modulation peaks
are equal and occur at the same percentage typical of conventional AM
modulation. I did not attempt to achieve more than 100% positive
modulation.
Distortion: At <100% modulation (95% typical): < 1% THD+n at 400 Hz., < 3% THD+n at 40 Hz and 10 KHz.
Carrier noise:
Bandwidth limited 50 Hz - 7.5 KHz is some -55 dB below 100%
modulation. This was measured at the cathode of the
6DE7. Demodulated RF waveform confirmed these values.
*
This value seemed to vary noticeably with various 6146A
tubes. Some tubes were good to 14 kHz while others seemed to
start a roll-off in the 7 to 10 kHz range.
Upper
end frequency response appears to be limited by the screen bypass C
(pin 3, 6146A) and by the coupling values of C and grid R's on the
lower end. There is a good bit of LF flicker noise generated at
the input 12AX7 stage. Efforts to push the low end down further
in frequency by further increasing coupling R’s greatly adds to
flicker noise. Perhaps a 7025 in place of the 12AX7 might solve
this.
The Mods/ Details
The following table shows the original values and the new
values where applicable.
|
Component
|
Original Value
|
New Value
|
Comments
|
| R22 |
1 Meg |
1 Meg |
|
| R23 |
4700 |
0.05 uF |
|
| R24 |
2.2 Meg |
2.2 Meg |
|
| R25 |
470 k |
470 k |
|
| R26 |
100 k |
100 k |
Might be 500 k in some TX's which is OK. |
| R27 |
470 k |
470 k |
|
| R28 |
3300 |
3300 |
|
| R29 |
2.2 Meg |
Removed |
|
| R30 |
1 Meg |
220 k 2 W |
|
| R31 |
33 k |
12 K 10 W |
Caution: The value has a major effect on positive modulation. Lower values
increases dissipation of the 6DE7 triode. |
| R32 |
10 k |
Removed |
|
| R33 |
33 k |
33 k |
|
| C29 |
100 pf |
100 pf |
|
| C30 |
0.001 uF |
Removed |
|
| C31 |
0.005 uF |
0.05 mF |
|
| C32 |
20uF@350V |
20uF@350V |
|
| C33 |
0.005 uF |
0.05 uF |
|
| C34 |
100 pf |
Removed |
|
| C35 |
0.1 uF |
Removed |
|
| C42 |
0.005 uF |
0.001 uF |
Screen bypass |
| C43 |
0.001 uF |
Removed |
|
Table 1 - Component Info
See the schematic
for the changes and values of new added components. Modifications
to the transmitter are strictly wiring and component changes with a
couple of terminal strips added using existing screws at the tube
sockets. No holes and no added things to deface or detract from the
value of the DX-60 were required. The zener value can be
made up of anything available to arrive at a reference of 120 V.
In my case I used a 100 V and a 20V in series. A pair of 60 V _
watt zeners will work as well.
The
PA operating parameters of the 6146A in CW tune are 650 VDC at 175 mA
and 2.5 mA of grid current. In AM they are 700 VDC at 70 mA (no
modulation) and 2.8 mA of grid current. The Plate meter just twitches
upward on modulation. All in all this is pure screen modulation
with PA efficiency at 68% thus about what is expected with a class C
amp.
Voltage values DCV: AM mode, no signal
|
12AX7
|
| Pin 1 |
138
|
| Pin 2 |
0
|
| Pin 3 |
1.28
|
| Pin 6 |
151
|
| Pin 7 |
0
|
| Pin 8 |
1.42
|
|
6DE7
|
| Pin 1 |
650
|
| Pin 2, 3 |
-31
|
| Pin 6 |
300
|
| Pin 7 |
0
|
| Pin 8 |
11.6
|
| Pin 9 |
68
|
Table 2 - Voltage Info
|