What began as an improved voltage regulator for the Phonoclone phono stage has morphed into a full circuit in it's own right. I thought I might post it to the power supply design forum, for those interested.
Full details are available here:
http://www.geocities.com/rjm003.geo/rjmaudio/diy_xreg.html
The development history can be traced via the Phonoclone 3 thread over at the analog forum:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=133191&perpage=25&pagenumber=1
So what's the Xreg?
Replaces LM7812, LM7912 or similar pair of voltage regulators that typically power op amp audio circuits. It's built from a pair of high performance op amps and complimentary pass transistors. What is (relatively) novel is that the op amps are configured to run in inverting mode, and the voltage references are generated by passive RCRC filters. You should be able to use this circuit board (downloadable for free on the RJM Audio website) to upgrade most preamps, phonostages, and headphone amplifiers that require a split supply.
I've designed it, built it, and listened to it, and it seems to work pretty darned well. I'm always up for constructive criticism and comments though.
-rjm
Full details are available here:
http://www.geocities.com/rjm003.geo/rjmaudio/diy_xreg.html
The development history can be traced via the Phonoclone 3 thread over at the analog forum:
http://www.diyaudio.com/forums/showthread.php?s=&threadid=133191&perpage=25&pagenumber=1
So what's the Xreg?
Replaces LM7812, LM7912 or similar pair of voltage regulators that typically power op amp audio circuits. It's built from a pair of high performance op amps and complimentary pass transistors. What is (relatively) novel is that the op amps are configured to run in inverting mode, and the voltage references are generated by passive RCRC filters. You should be able to use this circuit board (downloadable for free on the RJM Audio website) to upgrade most preamps, phonostages, and headphone amplifiers that require a split supply.
I've designed it, built it, and listened to it, and it seems to work pretty darned well. I'm always up for constructive criticism and comments though.
-rjm
Attachments
Specifically, I worry about the consequences of putting the pass transistor inside the feedback loop, with a moderately large reserve of capacitance on the output.
I don't really understand how things like op amp slew rate and transistor gain come into play, how difficult the transistor base is to drive at high frequencies, how capable the op amp is of doing so, and how much more difficult this is made by the presence of significant output capacitance.
The fact that the sound changed quite dramatically when 47 ohm resistors were inserted between the op amp output and transistor base suggest to me that things are more complicated in this regard than they first appear. I am looking for advice on the matter.
I don't really understand how things like op amp slew rate and transistor gain come into play, how difficult the transistor base is to drive at high frequencies, how capable the op amp is of doing so, and how much more difficult this is made by the presence of significant output capacitance.
The fact that the sound changed quite dramatically when 47 ohm resistors were inserted between the op amp output and transistor base suggest to me that things are more complicated in this regard than they first appear. I am looking for advice on the matter.
Progress update:
There is a single-sided board for the Xreg now, using a dual op-amp.
This makes it a lot simpler to built, point-to-point is doable using essentially the same layout as the printed circuit board.
Eagle files can be downloaded from the Xreg web page.
/rjm
There is a single-sided board for the Xreg now, using a dual op-amp.
This makes it a lot simpler to built, point-to-point is doable using essentially the same layout as the printed circuit board.
Eagle files can be downloaded from the Xreg web page.
/rjm
What you have there is a "capacitance multiplier" rather than a voltage regulator. The difference being that the reference voltage is derived from an RC (or RCRC in your case) filter instead of a fixed voltage reference. It behaves similarly to a huge capacitor, as implied by the name.
There should be no particular problem with including the pass transistor in the feedback loop. Try running the circuit through a simulator and seeing what the phase margin is likely to be. If it is poor, you may need to limit the bandwidth with a small capacitor from the op-amp's output to its inverting input.
To test the real circuit, try putting a large capacitance across the output, then loading it with a square wave current, e.g. a power amp playing a square wave into a resistive load. Use an oscilloscope to check for ringing on the output (ideally there will be none).
However, you will find that there is no benefit from adding any significant capacitance to the output.
There should be no particular problem with including the pass transistor in the feedback loop. Try running the circuit through a simulator and seeing what the phase margin is likely to be. If it is poor, you may need to limit the bandwidth with a small capacitor from the op-amp's output to its inverting input.
To test the real circuit, try putting a large capacitance across the output, then loading it with a square wave current, e.g. a power amp playing a square wave into a resistive load. Use an oscilloscope to check for ringing on the output (ideally there will be none).
However, you will find that there is no benefit from adding any significant capacitance to the output.
Insofar as the Xreg acts like a capacitor I can appreciate that additional physical output capacitance is of no benefit.
Since the op amp circuit it powers normally has about 100 uF of bypass capacitance anyway, are C6(A) really needed? Probably not, but as a stability measure it probably doesn't hurt either.
I will try to do some square wave testing when I get the chance. Since the original post I've done extensive listening tests and it all seems to work quite well, but I would like to have some measurements to back that up.
Since the op amp circuit it powers normally has about 100 uF of bypass capacitance anyway, are C6(A) really needed? Probably not, but as a stability measure it probably doesn't hurt either.
I will try to do some square wave testing when I get the chance. Since the original post I've done extensive listening tests and it all seems to work quite well, but I would like to have some measurements to back that up.
You should have at least some capacitance on the output for stability. You want enough to keep the output impedance low at higher frequencies where the feedback starts to drop. I reckon about 1uF would be enough for that circuit, since it should have pretty low output impedance already, but the only way to be sure is to test it.
Hi RJM,
First time I've looked properly at the Xreg (linked from your preamp thread) - I remember reading about your development of it with the Phonoclone 3, but I've spent a year studying since then, so now what it does makes sense. Very elegant, I like the way you're not fussed about the absolute value of the rails.
I've been looking for a nice design to use with a headphone amp - it uses a virtual ground, though, as built - does it make more sense to:
a) star ground both the Xreg's COM and the headamp's Vground to chassis (but otherwise floating) OR
b) ground the center tap at the transformer, and remove the TLE2426 from the headphone amp in favour of a real ground?
First time I've looked properly at the Xreg (linked from your preamp thread) - I remember reading about your development of it with the Phonoclone 3, but I've spent a year studying since then, so now what it does makes sense. Very elegant, I like the way you're not fussed about the absolute value of the rails.
I've been looking for a nice design to use with a headphone amp - it uses a virtual ground, though, as built - does it make more sense to:
a) star ground both the Xreg's COM and the headamp's Vground to chassis (but otherwise floating) OR
b) ground the center tap at the transformer, and remove the TLE2426 from the headphone amp in favour of a real ground?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.