331
Equipment / What's all this regenerative loop stuff?
« on: April 20, 2018, 1311 UTC »
I have just verified the operation of a resonant NDB loop which uses positive feedback to enhance the Q which can be attained with commonly available real world components, namely, tuning diodes and Litz-wound ferrite rods. A working prototype is up and running in the basement here and is showing great promise.
I have re-tooled my ferrite loop so that it is now tuned remotely by a pair of back to back NTE618 tuning diodes (4 diodes total) via a control signal of 0 - 15 VDC. These diodes resonate a wound ferrite rod (160 turns of Litz) over the NDB band. The loop output is amplified by a balanced JFET differential amp. Tuning diodes work great with one caveat: they have a Q which increases with frequency, and vicey-versey. At the top end of the band (around 400 kHz) they have enough Q to produce a reasonable but not great loaded Q of around 180. I would really like this to be higher; Q = 250 would be acceptable . At the low end of the band (around 200 kHz) the loaded Q is only around 90, and the loop, well, the loop sucks. There is just too much loss due to series resistance in the loop tank, and this needs to be reduced for the loop to perform as desired. One could use lower loss components in the tank, but this is not practical. Instead, one can trick the loop into believing that it has lower losses. If one replaces some of the energy burned up in the lossy tuning diodes and ferrite rod with some energy derived from the loop amp output, then the loop Q should increase. The loop now becomes a regenerative loop which employs positive feedback for Q augmentation.
The best way to see this is through an understanding of what Q represents. We have all probably seen that the Q of a resonant loop is the ratio of the loop's resonant frequency to the width of the response curve between its -3 dB points. As the Q is increased, the loop tunes more sharply, and its -3 dB points move in towards the center frequency. This is really a result of Q and is not a formal definition of Q. In electrical engineering, the Q of an LC circuit is defined as the ratio of the energy stored in the L and the C per cycle to the energy dissipated in losses per cycles. If one can replace some or all of the losses dissipated as heat, then the loop Q must increase. As one adds controlled positive feedback to the LC tank, one is in effect adding negative resistance to the loss elements in the LC tank.
In practice, I take the output of the loop amp and split it using a toroidal splitter. One port goes to the coax back to the SDR. The other port is attenuated in a 75 ohm 20 dB attenuator and passed to a variable attenuator comprised of a PIN diode and load resistor. This voltage is then applied to a one turn tickler winding on the ferrite rod. The biasing of the PIN diode is set up so that a 0.7 - 5V control signal will cause a current of 0 to 150 uA to flow through the PIN diode, thus changing its effective AC series resistance.
I have observed 3 regions of operation. Below a PIN diode current of 1 uA, the loop behaves like the underlying loop without regeneration, and has a generally depressed Q due to the Q of the NTE618 diodes. If the PIN diode current is increased to a value of typically 100 uA, the loop breaks into oscillation (as it should) and the SDR waterfall goes bat-guano crazy. Between these 2 points, one can observe the loop Q increasing smoothly above the default level with increasing PIN diode current. The response in the spectrum pane of the SDR begins to peak higher, and the response skirts narrow. The object viewed near the resonant peak (carrier or offset) in the waterfall begins to brighten with increasing PIN diode current. I can typically see an increase of up to 20 dBm in this middle region with increasing PIN diode current before it breaks into oscillation.
The regeneration process seems to be well behaved and stable as the Q operating point is varied, and this bodes well for the future. I am not anticipating running the loop at insanely high Qs such as 10000; I simply would like to enhance the Q of a deficient loop to around 250 which experience has shown is about what is required to make the loop competitive with the amplified whip which I also employ.
Interesting sidebar: if the sense of the tickler winding is reversed - negative feedback - then the regeneration will produce increasing attenuation of received signals as the PIN diode current is increased.
What I now have is a resonant NDB loop with software programmable (and knobless) control of:
- The loop azimuthal position, with an accuracy of better than 1 degree.
- The loop resonant frequency with a resolution of typically 100 Hz.
- The loop Q and thus its Effective Height.
Further details will be posted after it is sited outdoors and fully shaken down.
I have re-tooled my ferrite loop so that it is now tuned remotely by a pair of back to back NTE618 tuning diodes (4 diodes total) via a control signal of 0 - 15 VDC. These diodes resonate a wound ferrite rod (160 turns of Litz) over the NDB band. The loop output is amplified by a balanced JFET differential amp. Tuning diodes work great with one caveat: they have a Q which increases with frequency, and vicey-versey. At the top end of the band (around 400 kHz) they have enough Q to produce a reasonable but not great loaded Q of around 180. I would really like this to be higher; Q = 250 would be acceptable . At the low end of the band (around 200 kHz) the loaded Q is only around 90, and the loop, well, the loop sucks. There is just too much loss due to series resistance in the loop tank, and this needs to be reduced for the loop to perform as desired. One could use lower loss components in the tank, but this is not practical. Instead, one can trick the loop into believing that it has lower losses. If one replaces some of the energy burned up in the lossy tuning diodes and ferrite rod with some energy derived from the loop amp output, then the loop Q should increase. The loop now becomes a regenerative loop which employs positive feedback for Q augmentation.
The best way to see this is through an understanding of what Q represents. We have all probably seen that the Q of a resonant loop is the ratio of the loop's resonant frequency to the width of the response curve between its -3 dB points. As the Q is increased, the loop tunes more sharply, and its -3 dB points move in towards the center frequency. This is really a result of Q and is not a formal definition of Q. In electrical engineering, the Q of an LC circuit is defined as the ratio of the energy stored in the L and the C per cycle to the energy dissipated in losses per cycles. If one can replace some or all of the losses dissipated as heat, then the loop Q must increase. As one adds controlled positive feedback to the LC tank, one is in effect adding negative resistance to the loss elements in the LC tank.
In practice, I take the output of the loop amp and split it using a toroidal splitter. One port goes to the coax back to the SDR. The other port is attenuated in a 75 ohm 20 dB attenuator and passed to a variable attenuator comprised of a PIN diode and load resistor. This voltage is then applied to a one turn tickler winding on the ferrite rod. The biasing of the PIN diode is set up so that a 0.7 - 5V control signal will cause a current of 0 to 150 uA to flow through the PIN diode, thus changing its effective AC series resistance.
I have observed 3 regions of operation. Below a PIN diode current of 1 uA, the loop behaves like the underlying loop without regeneration, and has a generally depressed Q due to the Q of the NTE618 diodes. If the PIN diode current is increased to a value of typically 100 uA, the loop breaks into oscillation (as it should) and the SDR waterfall goes bat-guano crazy. Between these 2 points, one can observe the loop Q increasing smoothly above the default level with increasing PIN diode current. The response in the spectrum pane of the SDR begins to peak higher, and the response skirts narrow. The object viewed near the resonant peak (carrier or offset) in the waterfall begins to brighten with increasing PIN diode current. I can typically see an increase of up to 20 dBm in this middle region with increasing PIN diode current before it breaks into oscillation.
The regeneration process seems to be well behaved and stable as the Q operating point is varied, and this bodes well for the future. I am not anticipating running the loop at insanely high Qs such as 10000; I simply would like to enhance the Q of a deficient loop to around 250 which experience has shown is about what is required to make the loop competitive with the amplified whip which I also employ.
Interesting sidebar: if the sense of the tickler winding is reversed - negative feedback - then the regeneration will produce increasing attenuation of received signals as the PIN diode current is increased.
What I now have is a resonant NDB loop with software programmable (and knobless) control of:
- The loop azimuthal position, with an accuracy of better than 1 degree.
- The loop resonant frequency with a resolution of typically 100 Hz.
- The loop Q and thus its Effective Height.
Further details will be posted after it is sited outdoors and fully shaken down.