Part of the ongoing saga of an amp modification -- needed to reduce Vcc/Vee to +/-20VDC to run the OPA2604's -- I had done some work on morgan jones "two transistor" but it was getting to be a little bit more of a furball than i wanted. Took a look at the original 1980 circuit and some modes of output compensation -- they used a 1u/2.7R Zobel on the output -- and I didn't think it would be stable with this combination:
An externally hosted image should be here but it was not working when we last tested it.
In the top panel I prefer the blue & yellow traces; I like the rolloff to keep rolling off so there's plenty of gain margin.
In the bottom panel I prefer the blue & red traces because they have the greatest phase margin (if I'm reading the diagram correctly).
So I guess that of these four candidates I prefer the blue.
But why do a Maida at all? In the year of our lord 2013 we've got cheap and rugged HEXFETs; why not build a straightforward MOSFET "preregulator", with zener diode setting the gate voltage, that outputs 24V? Then follow with a standard LM317 regulator having +24V input and +20V output. Let the first stage do the voltage-dropping and let the second stage do the voltage-regulating.
In the bottom panel I prefer the blue & red traces because they have the greatest phase margin (if I'm reading the diagram correctly).
So I guess that of these four candidates I prefer the blue.
But why do a Maida at all? In the year of our lord 2013 we've got cheap and rugged HEXFETs; why not build a straightforward MOSFET "preregulator", with zener diode setting the gate voltage, that outputs 24V? Then follow with a standard LM317 regulator having +24V input and +20V output. Let the first stage do the voltage-dropping and let the second stage do the voltage-regulating.
Last edited:
Here's another tidbit -- i saw it in simulation, but would have to verify by running it on the analyzer --
An externally hosted image should be here but it was not working when we last tested it.
Here's a non-Maida circuit that uses an Nchannel power MOSFET to drop the voltage. It doesn't bootstrap the first stage input from the second stage's output, so the feedback loop is drastically simpler than in the Maida configuration.
If you so choose, you can set a 6 Hz pole at the (hi-Z!) MOSFET gate node, without any fear of destabilizing the LM317. A pole at 6 Hz improves line regulation by another 20 dB at the 120 Hz line ripple frequency.
It's not especially sensitive to parts choice, although I do recommend using a "Logic Level Gate" MOSFET just to reduce the amount of uncertainty in VGS. Part numbers shown were the cheapest I found on Digikey @ qty_1, having what I consider "sensible safety margins" and what you may consider "ridiculous overkill".
If you so choose, you can set a 6 Hz pole at the (hi-Z!) MOSFET gate node, without any fear of destabilizing the LM317. A pole at 6 Hz improves line regulation by another 20 dB at the 120 Hz line ripple frequency.
It's not especially sensitive to parts choice, although I do recommend using a "Logic Level Gate" MOSFET just to reduce the amount of uncertainty in VGS. Part numbers shown were the cheapest I found on Digikey @ qty_1, having what I consider "sensible safety margins" and what you may consider "ridiculous overkill".
Attachments
- Status
- Not open for further replies.