11-07-2020, 06:43 PM
(This post was last modified: 11-07-2020, 06:46 PM by Mike Watterson.)
Dunno.
I did look at the single diode detector to 0V and found that it doesn't work properly without a large series resistor between the diode and an RF decoupling cap. Because otherwise with no RF decoulper there is RF on the grid, and with one the previous stage is loaded badly. If you look at actual valve radio circuits you'll see a resistor between the diode anode and the volume control, which has the RF decoupling on it. Similarly an AGC line has a series resistor.
So I then tried the peak detector. The first diode can be like on a valve radio, cathode to 0V. Then the next diode, D2, has the cathode to D1's anode. The grid leak and RF decoupling are then the load on the Anode of D2. A peak detector is usually drawn with the diodes the other way round. If the g1a and g1b are separate but connected by a 10nF cap, and g1b has 2.2M to 10 M Grid leak, and diodes fed DC to g1a with a same value grid leak, then at low signals the g1a and g1b have same voltage, but as signal gets larger the g1a is more negative and the gain drops. I've not tried it yet. Maybe tomorrow.
However EITHER you use one diode and need a series resistor before the RF decoupling, OR you use two diodes wired as a peak detector. Analysis will show that the load resistance is the AC load, but that because there are two diodes, the RF decoupler across the load resistor (the grid leak) is actually invisible to RF. The series capacitor from the previous stage only needs to be large enough for the equivalent RF AC load of the load resistor. The capacitor on the load resistor only needs to be small enough to not too much attenuate the audio given the Grid Leak value. At an IF/RF of 485 kHz, 3M3 Ohms load with 10M scope probe (10:1 into a 1M || 30pF scope) an RF decoupler of 680 pF was enough. 100% RF without it.
No extra series resistor is needed with the peak detector, so in practice the audio level is about 2.3x to 2.5x a single diode with the series resistor to feed the RF decoupling, depending on the source impedance and the final load resistor (the grid leak).
The capacitor to feed the peak detector was 220pF from the source coil.
An IFT was simulated by having 2 x 100 uH "resistor body style" inductors on either side of the breadboard divide that's for down the middle of DIL sockets. These were tuned with 1.22nF. An old Advance signal gen fed one side via 1K, though even at direct (50? 75? what?) connection the peak was still fairly sharp.
I can't see how the RF valve wouldn't have been damaged.
I did look at the single diode detector to 0V and found that it doesn't work properly without a large series resistor between the diode and an RF decoupling cap. Because otherwise with no RF decoulper there is RF on the grid, and with one the previous stage is loaded badly. If you look at actual valve radio circuits you'll see a resistor between the diode anode and the volume control, which has the RF decoupling on it. Similarly an AGC line has a series resistor.
So I then tried the peak detector. The first diode can be like on a valve radio, cathode to 0V. Then the next diode, D2, has the cathode to D1's anode. The grid leak and RF decoupling are then the load on the Anode of D2. A peak detector is usually drawn with the diodes the other way round. If the g1a and g1b are separate but connected by a 10nF cap, and g1b has 2.2M to 10 M Grid leak, and diodes fed DC to g1a with a same value grid leak, then at low signals the g1a and g1b have same voltage, but as signal gets larger the g1a is more negative and the gain drops. I've not tried it yet. Maybe tomorrow.
However EITHER you use one diode and need a series resistor before the RF decoupling, OR you use two diodes wired as a peak detector. Analysis will show that the load resistance is the AC load, but that because there are two diodes, the RF decoupler across the load resistor (the grid leak) is actually invisible to RF. The series capacitor from the previous stage only needs to be large enough for the equivalent RF AC load of the load resistor. The capacitor on the load resistor only needs to be small enough to not too much attenuate the audio given the Grid Leak value. At an IF/RF of 485 kHz, 3M3 Ohms load with 10M scope probe (10:1 into a 1M || 30pF scope) an RF decoupler of 680 pF was enough. 100% RF without it.
No extra series resistor is needed with the peak detector, so in practice the audio level is about 2.3x to 2.5x a single diode with the series resistor to feed the RF decoupling, depending on the source impedance and the final load resistor (the grid leak).
The capacitor to feed the peak detector was 220pF from the source coil.
An IFT was simulated by having 2 x 100 uH "resistor body style" inductors on either side of the breadboard divide that's for down the middle of DIL sockets. These were tuned with 1.22nF. An old Advance signal gen fed one side via 1K, though even at direct (50? 75? what?) connection the peak was still fairly sharp.
I can't see how the RF valve wouldn't have been damaged.
