Hi James
It exists for the Mini refering to its build and there are special points for it to read but for the general stuff it links back to the original Ref-D's guide. I attach it here again for you to read it. Also see that post which explains the Vref elements pre-test process for the Mini to bits: http://www.diyaudio.com/forums/power-supplies/261031-reflektor-d-builds-71.html#post4710267
It exists for the Mini refering to its build and there are special points for it to read but for the general stuff it links back to the original Ref-D's guide. I attach it here again for you to read it. Also see that post which explains the Vref elements pre-test process for the Mini to bits: http://www.diyaudio.com/forums/power-supplies/261031-reflektor-d-builds-71.html#post4710267
Stiff Salas with Negative Output Impedance (adjustable)
Greetings,
Not sure if this is the most appropriate thread for this (off-topic?).
If there is a better thread, please kindly advise.
I was recently pondering about creating a regulator with adjustable negative output impedance, so as to compensate for the voltage sag at increased current draw.
Initial fiddling (Spice, Soldered prototype) with basic LM317 and LM337 confirmed that the concept is defendable: An increase of output current resulting in an increase (or at least "no change" when optimized) of output voltage.
Encouraged by this, I came up with this version of a Duo-Salas Regulator.
Or maybe a Stiff Duo-Salas with adjustable Negative Output Impedance.
Does this make any sense to you guys?
A penny for your thoughts.
Cheers,
Ziggy.
Greetings,
Not sure if this is the most appropriate thread for this (off-topic?).
If there is a better thread, please kindly advise.
I was recently pondering about creating a regulator with adjustable negative output impedance, so as to compensate for the voltage sag at increased current draw.
Initial fiddling (Spice, Soldered prototype) with basic LM317 and LM337 confirmed that the concept is defendable: An increase of output current resulting in an increase (or at least "no change" when optimized) of output voltage.
Encouraged by this, I came up with this version of a Duo-Salas Regulator.
Or maybe a Stiff Duo-Salas with adjustable Negative Output Impedance.
Does this make any sense to you guys?
A penny for your thoughts.
Cheers,
Ziggy.
Attachments
Last edited:
... and this would be a slightly slimmer version, without the redundant overhead, but still with the adjustable negative impedance capability ...
Comments, anybody?
Is there some inherent flaw in this line of thinking
(apart the larger number of parts used)?
Comments, anybody?
Is there some inherent flaw in this line of thinking
(apart the larger number of parts used)?
Attachments
I think that will try to test this following variation. This time there are two sets of capacitors. But first, let's imagine an imaginary "mid-height reference" level, a "mid-ground" of sorts, cutting the schematic horizontally at half height. Any increased current draw spike originating from the load (R_load which is also cut in half by this "imaginary" mid-height reference) could result in:
1). a slightly falling spike of the positive voltage of the upper rail, and
2). a slightly increasing spike (a decrease of negative voltage => increase) of the lower rail, assuming that they are referenced to the imaginary "mid-height" reference.
This rising spike of the lower rail is conveyed by the lower set of (speed-up) capacitors to the base of the upper control transistor. This (imho) causes that this transistor closes up quicker and "harder" than as compared to / under normal conditions, as when its emitter sees only the falling voltage spike of the positive rail.
Vice versa,
The falling spike of the upper rail is conveyed by the upper set of (speed-up) capacitors to the base of the lower control transistor. This also (ihmo) causes that it closes up quicker and "harder" than as compared to / under normal conditions, as when its emitter only sees the rising voltage spike of the negative rail.
In essence, these sets of speed up capacitors should provide a measure of additional negative impedance "boost" under AC conditions.
In terms of DC conditions, The key player is and remains the central potentiometer (R106), the one that defines the output voltage level. Each of the current sources, both above as well as below this potentiometer, sees a slightly lesser value of the set resistance, due to the fact that there are two thieves. Current thieves. The R8 and R108 which are stealing part of the current from the sources, diverting it from R106 via an alternative path to the opposite rail.
Therefore, in the case of a long term variation of voltage at the output (say, voltage sag due to increased current draw by load), the amount of current that is being "stolen" by the thieves (R8, and R108) becomes lesser.
In essence, the current sources are hence "fooled" to believe that someone came and increased the resistance of the R106 potentiometer, effectively increasing the output voltage, i.e. compensating (or even overcompensating) for the hypothetical voltage sag that was experienced at the output in the first place.
Mind you, these are all just preliminary assumptions, that would need some verification in simulation and praxis. Question remains - if indeed, adjustable negative output impedance can be obtained in this way, does the anticipated functionality offset and outweigh the increased count of elements and complexity. Not to mention the other question, a question of stability.
Negative impedance sounds like an oscillations-accident awaiting to happen. For a person which is not equipped with a scope, it would be rather difficult to hand tweak the settings of the upper and lower pots, so as to achieve an optimum point. This would render the concept potentially useless for DIY'ers who are not equipped with a scope.
Comments kindly welcome. These are all early stage speculations, pertaining to a fresh concept, so maybe there is a flaw somewhere, but I do not see it yet.
1). a slightly falling spike of the positive voltage of the upper rail, and
2). a slightly increasing spike (a decrease of negative voltage => increase) of the lower rail, assuming that they are referenced to the imaginary "mid-height" reference.
This rising spike of the lower rail is conveyed by the lower set of (speed-up) capacitors to the base of the upper control transistor. This (imho) causes that this transistor closes up quicker and "harder" than as compared to / under normal conditions, as when its emitter sees only the falling voltage spike of the positive rail.
Vice versa,
The falling spike of the upper rail is conveyed by the upper set of (speed-up) capacitors to the base of the lower control transistor. This also (ihmo) causes that it closes up quicker and "harder" than as compared to / under normal conditions, as when its emitter only sees the rising voltage spike of the negative rail.
In essence, these sets of speed up capacitors should provide a measure of additional negative impedance "boost" under AC conditions.
In terms of DC conditions, The key player is and remains the central potentiometer (R106), the one that defines the output voltage level. Each of the current sources, both above as well as below this potentiometer, sees a slightly lesser value of the set resistance, due to the fact that there are two thieves. Current thieves. The R8 and R108 which are stealing part of the current from the sources, diverting it from R106 via an alternative path to the opposite rail.
Therefore, in the case of a long term variation of voltage at the output (say, voltage sag due to increased current draw by load), the amount of current that is being "stolen" by the thieves (R8, and R108) becomes lesser.
In essence, the current sources are hence "fooled" to believe that someone came and increased the resistance of the R106 potentiometer, effectively increasing the output voltage, i.e. compensating (or even overcompensating) for the hypothetical voltage sag that was experienced at the output in the first place.
Mind you, these are all just preliminary assumptions, that would need some verification in simulation and praxis. Question remains - if indeed, adjustable negative output impedance can be obtained in this way, does the anticipated functionality offset and outweigh the increased count of elements and complexity. Not to mention the other question, a question of stability.
Negative impedance sounds like an oscillations-accident awaiting to happen. For a person which is not equipped with a scope, it would be rather difficult to hand tweak the settings of the upper and lower pots, so as to achieve an optimum point. This would render the concept potentially useless for DIY'ers who are not equipped with a scope.
Comments kindly welcome. These are all early stage speculations, pertaining to a fresh concept, so maybe there is a flaw somewhere, but I do not see it yet.
Attachments
Last edited:
9V Tx should allow about 6.5V drop across the reg which is enough. 12V Tx should allow about 10.5V across the reg which is nice from less CCS Mosfet parasitic capacitance point of view but that Mosfet's dissipation will be 60% more i.e. 3.25W vs 5.25W. Its firstly a matter of sinking and ventilation capability in a specific build. Maybe 9V Tx and 540mA CCS is a good compromise.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.